Spaces:

stevafernandes
/

cellquant

Sleeping

App Files Files Community

stevafernandes commited on 27 days ago

Commit

812b6ee

verified ·

1 Parent(s): 5d31c67

Deploy CellQuant

Browse files

Files changed (23) hide show

.gitattributes +0 -1
README.md +24 -17
cellquant/__init__.py +1 -1
cellquant/data.py +63 -9
cellquant/infer.py +17 -5
cellquant/losses.py +35 -8
cellquant/model.py +10 -1
cellquant/postprocess.py +20 -9
reports/SUMMARY.md +68 -35
reports/cv_v2/comparison.md +94 -0
reports/cv_v2/cv_report.md +105 -0
reports/cv_v2/cv_results.json +0 -0
reports/cv_v2/run_all.sh +11 -0
reports/cv_v2/train_queue.sh +11 -0
reports/final/final_fit.json +848 -848
scripts/build_summary.py +55 -9
scripts/compare_attributes.py +113 -0
scripts/compare_cv.py +110 -0
tests/test_data.py +3 -2
tests/test_v2_recipe.py +152 -0
train.py +26 -6
weights/cellquant.pt +2 -2
weights/cellquant_history.json +540 -480

.gitattributes CHANGED Viewed

	@@ -1,2 +1 @@
1	*.pt filter=lfs diff=lfs merge=lfs -text
2	- examples/F100_crop_512.png filter=lfs diff=lfs merge=lfs -text


1	*.pt filter=lfs diff=lfs merge=lfs -text

README.md CHANGED Viewed

@@ -32,10 +32,13 @@ Every nucleus receives exactly one of the four *Cell* labels.
 * **Model** – a UNet with an ImageNet-pretrained ResNet34 encoder (`cellquant/model.py`) that
   predicts, per pixel: nucleus interior / border / background, P(antibody +), P(nucleus has
-  punctum) and punctum in-nucleus / external / background. A full-resolution refinement stage
-  keeps 5-pixel puncta.
-* **Instances** – a marker-controlled watershed on the interior/border maps separates touching
-  nuclei; puncta are connected components (`cellquant/postprocess.py`).
 * **Labels** – antibody status is the mean probability inside each nucleus; a nucleus is
   "Nostril" when a punctum classified "in nucleus" lies inside it (the annotation protocol;
   the network's own nucleus-level head is available as an alternative in *Advanced*);
@@ -43,7 +46,7 @@ Every nucleus receives exactly one of the four *Cell* labels.
 * **Training data** – nine 1024×1024 RGB images with 1,569 nucleus outlines and 1,057
   punctum outlines, parsed directly from the CVAT zip exports (`cellquant/annotations.py`).
-## Validation (see `reports/SUMMARY.md` and `reports/cv/cv_report.md`)
 Leave-one-image-out cross-validation: nine models, each tested on an image it never saw, and
 each image tested 23 times (repeat runs, TTA settings, 8 rotations/flips, 6 photometric
@@ -51,17 +54,21 @@ perturbations, rule variants). Micro-averaged over the nine held-out images:
 | Measure | Precision | Recall | F1 |
 |---|---:|---:|---:|
-| Nucleus detection, IoU ≥ 0.5 | 77.5% | 78.4% | 77.9% |
-| Nucleus detection, centroid inside the annotated nucleus | 87.0% | 88.0% | 87.5% |
-| Punctum detection (centroid within 6 px) | 77.3% | 82.8% | 79.9% |
-Among matched nuclei the antibody +/− call is right for 94% and the Nostril attribute for
-96%; puncta are classified in-nucleus vs external correctly 93% of the time. Scores vary by
-at most ±2 F1 points across the 14 rotated/perturbed runs of each image, and repeated runs are
-bit-identical. The weak spots are antibody-negative nuclei (F1 32%; only 102 examples, 62 of
-them in one image), the five-example "Cell, Nostril, Antibody −" class, and clusters of touching
-nuclei that are sometimes merged. The shipped `weights/cellquant.pt` was trained on all nine
-images with the same recipe.
 ## Using the Space
@@ -104,7 +111,7 @@ Retrain when more annotated images are available (put the TIFs and the CVAT zips
 ```bash
 python train.py --data-dir /path/to/folder --out weights/cellquant.pt --iters 3000
-python evaluate.py cv --data-dir /path/to/folder --runs-dir runs/cv --iters 3000
 ```
 Training uses Apple-silicon MPS, CUDA or CPU automatically.

 * **Model** – a UNet with an ImageNet-pretrained ResNet34 encoder (`cellquant/model.py`) that
   predicts, per pixel: nucleus interior / border / background, P(antibody +), P(nucleus has
+  punctum), punctum in-nucleus / external / background, and the normalised distance to the
+  nucleus outline (1 at the centre of every nucleus, 0 at its outline). A full-resolution
+  refinement stage keeps 5-pixel puncta.
+* **Instances** – nucleus foreground comes from the class map; markers (distance > 0.5) and the
+  watershed landscape come from the predicted distance map, which has one peak per nucleus and
+  a valley between touching ones; puncta are connected components (`cellquant/postprocess.py`).
+  Checkpoints without the distance channel fall back to a watershed on the interior map.
 * **Labels** – antibody status is the mean probability inside each nucleus; a nucleus is
   "Nostril" when a punctum classified "in nucleus" lies inside it (the annotation protocol;
   the network's own nucleus-level head is available as an alternative in *Advanced*);
 * **Training data** – nine 1024×1024 RGB images with 1,569 nucleus outlines and 1,057
   punctum outlines, parsed directly from the CVAT zip exports (`cellquant/annotations.py`).
+## Validation (see `reports/SUMMARY.md`, `reports/cv_v2/cv_report.md` and `reports/cv_v2/comparison.md`)
 Leave-one-image-out cross-validation: nine models, each tested on an image it never saw, and
 each image tested 23 times (repeat runs, TTA settings, 8 rotations/flips, 6 photometric
 | Measure | Precision | Recall | F1 |
 |---|---:|---:|---:|
+| Nucleus detection, IoU ≥ 0.5 | 75.8% | 85.5% | 80.4% |
+| Nucleus detection, centroid inside the annotated nucleus | 83.7% | 94.5% | 88.8% |
+| Punctum detection (centroid within 6 px) | 76.4% | 84.8% | 80.4% |
+Among matched nuclei the antibody +/− call is right for 93.5% and the Nostril attribute for
+95.6%; puncta are classified in-nucleus vs external correctly 93.4% of the time. Scores vary by
+at most ±3 F1 points across the 14 rotated/perturbed runs of each image, and repeated runs are
+bit-identical. The weak spots are antibody-negative nuclei (F1 38%; only 110 examples, 62 of
+them in one image, and the class-balanced loss trades their precision for recall), the five-example
+"Cell, Nostril, Antibody −" class, dense clusters of touching nuclei (still the main source of misses,
+although the distance-map watershed of this version recovers many that the first version merged), and
+dim nuclei the annotator skipped, which the model tends to count (about 13 % more nuclei than annotated).
+The previous version (v1: 2-px border, watershed on the interior map, unweighted antibody loss) scored
+77.9% nucleus F1, 87.5% by the centroid rule and 79.9% puncta F1 on the same protocol; `reports/cv/`
+keeps its report. The shipped `weights/cellquant.pt` was trained on all nine images with the v2 recipe.
 ## Using the Space
 ```bash
 python train.py --data-dir /path/to/folder --out weights/cellquant.pt --iters 3000
+python evaluate.py cv --data-dir /path/to/folder --runs-dir runs/cv_v2 --iters 3000
 ```
 Training uses Apple-silicon MPS, CUDA or CPU automatically.

cellquant/__init__.py CHANGED Viewed

@@ -3,4 +3,4 @@
 Trained on CVAT polygon annotations of 1024x1024 RGB images (blue = DAPI nuclei,
 green = antibody stain such as Iba1 or Tmem119, red = puncta).
 """
-__version__ = "1.0.0"

 Trained on CVAT polygon annotations of 1024x1024 RGB images (blue = DAPI nuclei,
 green = antibody stain such as Iba1 or Tmem119, red = puncta).
 """
+__version__ = "1.1.0"

cellquant/data.py CHANGED Viewed

@@ -6,6 +6,9 @@ Targets built from a :class:`RasterAnnotation` (all (H, W)):
     cellnos   float  1 where nucleus has a punctum ("Nostril"); valid on nucleus px
     cellmask  float  1 on nucleus pixels (loss mask for the two attribute heads)
     nos3      int64  0 bg, 1 punctum in nucleus, 2 punctum external
 """
 from __future__ import annotations
@@ -19,7 +22,35 @@ from skimage.segmentation import find_boundaries
 from .annotations import RasterAnnotation
-BORDER_WIDTH = 2  # pixels of the inner border ring
 def make_targets(R: RasterAnnotation) -> dict[str, np.ndarray]:
@@ -43,6 +74,7 @@ def make_targets(R: RasterAnnotation) -> dict[str, np.ndarray]:
         "cellnos": cn,
         "cellmask": fg.astype(np.float32),
         "nos3": nos3,
     }
@@ -60,6 +92,7 @@ class AugConfig:
     p_blur: float = 0.3
     p_channel_shift: float = 0.5   # per-channel additive offset
     offset_max: float = 0.03
 class PatchSampler:
@@ -67,33 +100,53 @@ class PatchSampler:
     Geometric: random scale (zoom), random crop, random dihedral transform.
     Photometric: per-channel gain, gamma, additive offset, Gaussian noise, blur.
-    Label maps use nearest-neighbour resampling.
     """
     def __init__(self, images: list[np.ndarray], targets: list[dict[str, np.ndarray]],
-                 cfg: AugConfig | None = None, seed: int = 0):
         assert len(images) == len(targets) and len(images) > 0
         self.images = images
         self.targets = targets
         self.cfg = cfg or AugConfig()
         self.rng = np.random.default_rng(seed)
     def _one(self) -> tuple[np.ndarray, dict[str, np.ndarray]]:
         cfg, rng = self.cfg, self.rng
-        i = int(rng.integers(len(self.images)))
         img, tg = self.images[i], self.targets[i]
         H, W = img.shape[:2]
         s = float(rng.uniform(cfg.scale_min, cfg.scale_max))      # zoom factor applied to the crop
         src = int(round(cfg.crop / s))                              # source window size
         src = max(32, min(src, H, W))
-        y0 = int(rng.integers(0, H - src + 1))
-        x0 = int(rng.integers(0, W - src + 1))
         patch = img[y0:y0 + src, x0:x0 + src].astype(np.float32) / 255.0
         tpatch = {k: v[y0:y0 + src, x0:x0 + src] for k, v in tg.items()}
         if src != cfg.crop:
             zoom = cfg.crop / src
             patch = ndi.zoom(patch, (zoom, zoom, 1), order=1)
-            tpatch = {k: ndi.zoom(v, (zoom, zoom), order=0) for k, v in tpatch.items()}
             patch = patch[:cfg.crop, :cfg.crop]
             tpatch = {k: v[:cfg.crop, :cfg.crop] for k, v in tpatch.items()}
             if patch.shape[0] != cfg.crop or patch.shape[1] != cfg.crop:  # rounding guard
@@ -145,7 +198,8 @@ class Prefetcher:
     fully determined by (seed, workers) regardless of thread scheduling.
     """
-    def __init__(self, images, targets, cfg: AugConfig, batch: int, seed: int = 0, workers: int = 3, depth: int = 6):
         import queue
         import threading
         workers = max(1, int(workers))
@@ -156,7 +210,7 @@ class Prefetcher:
         self.qs: list["queue.Queue"] = [queue.Queue(maxsize=max(1, depth // workers)) for _ in range(workers)]
         self._threads = []
         for w in range(workers):
-            s = PatchSampler(images, targets, cfg, seed=seed * 1000 + w)
             th = threading.Thread(target=self._work, args=(s, self.qs[w]), daemon=True)
             th.start()
             self._threads.append(th)

     cellnos   float  1 where nucleus has a punctum ("Nostril"); valid on nucleus px
     cellmask  float  1 on nucleus pixels (loss mask for the two attribute heads)
     nos3      int64  0 bg, 1 punctum in nucleus, 2 punctum external
+    dist      float  distance to the nucleus outline divided by the nucleus' maximum
+                     (1 at the centre, ~0 at the outline and between touching nuclei,
+                     0 outside); regression target of the distance channel
 """
 from __future__ import annotations
 from .annotations import RasterAnnotation
+BORDER_WIDTH = 3  # pixels of the inner border ring
+def distance_target(inst: np.ndarray) -> np.ndarray:
+    """Per-instance Euclidean distance to the nearest pixel outside the instance
+    (other instances and the image edge count as outside), divided by the
+    instance's maximum so that every nucleus peaks at exactly 1."""
+    out = np.zeros(inst.shape, dtype=np.float32)
+    for i, sl in enumerate(ndi.find_objects(inst), start=1):
+        if sl is None:
+            continue
+        m = inst[sl] == i
+        d = ndi.distance_transform_edt(np.pad(m, 1))[1:-1, 1:-1]
+        mx = float(d.max())
+        if mx > 0:
+            region = out[sl]
+            region[m] = (d / mx)[m]
+    return out
+def negative_nucleus_points(R: RasterAnnotation) -> np.ndarray:
+    """(N, 2) array of (y, x) centroids of the antibody-negative nuclei, used to
+    oversample crops that contain the rare class."""
+    neg = np.flatnonzero(~R.cell_antibody())
+    neg = neg[neg > 0]
+    if neg.size == 0:
+        return np.zeros((0, 2), dtype=np.float64)
+    ones = np.ones_like(R.cell_inst, dtype=np.float32)
+    return np.array(ndi.center_of_mass(ones, R.cell_inst, neg), dtype=np.float64).reshape(-1, 2)
 def make_targets(R: RasterAnnotation) -> dict[str, np.ndarray]:
         "cellnos": cn,
         "cellmask": fg.astype(np.float32),
         "nos3": nos3,
+        "dist": distance_target(inst),
     }
     p_blur: float = 0.3
     p_channel_shift: float = 0.5   # per-channel additive offset
     offset_max: float = 0.03
+    p_focus: float = 0.25          # share of crops forced to contain an antibody-negative nucleus
 class PatchSampler:
     Geometric: random scale (zoom), random crop, random dihedral transform.
     Photometric: per-channel gain, gamma, additive offset, Gaussian noise, blur.
+    Label maps use nearest-neighbour resampling (the distance map is bilinear).
+    ``focus``: optional per-image (N, 2) arrays of (y, x) points; a share
+    ``cfg.p_focus`` of the crops is placed so that it contains one such point,
+    choosing the image in proportion to its number of points (oversampling of
+    the antibody-negative nuclei).
     """
     def __init__(self, images: list[np.ndarray], targets: list[dict[str, np.ndarray]],
+                 cfg: AugConfig | None = None, seed: int = 0, focus: list[np.ndarray] | None = None):
         assert len(images) == len(targets) and len(images) > 0
         self.images = images
         self.targets = targets
         self.cfg = cfg or AugConfig()
         self.rng = np.random.default_rng(seed)
+        self.focus = focus
+        self._focus_images = [i for i, f in enumerate(focus or []) if f is not None and len(f)]
+        n = np.array([len(focus[i]) for i in self._focus_images], dtype=np.float64)
+        self._focus_p = n / n.sum() if n.size else n
     def _one(self) -> tuple[np.ndarray, dict[str, np.ndarray]]:
         cfg, rng = self.cfg, self.rng
+        point = None
+        if self._focus_images and rng.random() < cfg.p_focus:
+            i = self._focus_images[int(rng.choice(len(self._focus_images), p=self._focus_p))]
+            point = self.focus[i][int(rng.integers(len(self.focus[i])))]
+        else:
+            i = int(rng.integers(len(self.images)))
         img, tg = self.images[i], self.targets[i]
         H, W = img.shape[:2]
         s = float(rng.uniform(cfg.scale_min, cfg.scale_max))      # zoom factor applied to the crop
         src = int(round(cfg.crop / s))                              # source window size
         src = max(32, min(src, H, W))
+        if point is None:
+            y0 = int(rng.integers(0, H - src + 1))
+            x0 = int(rng.integers(0, W - src + 1))
+        else:  # any window that contains the point
+            py = min(H - 1, max(0, int(round(point[0]))))
+            px = min(W - 1, max(0, int(round(point[1]))))
+            y0 = int(rng.integers(max(0, py - src + 1), min(H - src, py) + 1))
+            x0 = int(rng.integers(max(0, px - src + 1), min(W - src, px) + 1))
         patch = img[y0:y0 + src, x0:x0 + src].astype(np.float32) / 255.0
         tpatch = {k: v[y0:y0 + src, x0:x0 + src] for k, v in tg.items()}
         if src != cfg.crop:
             zoom = cfg.crop / src
             patch = ndi.zoom(patch, (zoom, zoom, 1), order=1)
+            tpatch = {k: ndi.zoom(v, (zoom, zoom), order=1 if k == "dist" else 0) for k, v in tpatch.items()}
             patch = patch[:cfg.crop, :cfg.crop]
             tpatch = {k: v[:cfg.crop, :cfg.crop] for k, v in tpatch.items()}
             if patch.shape[0] != cfg.crop or patch.shape[1] != cfg.crop:  # rounding guard
     fully determined by (seed, workers) regardless of thread scheduling.
     """
+    def __init__(self, images, targets, cfg: AugConfig, batch: int, seed: int = 0, workers: int = 3, depth: int = 6,
+                 focus: list[np.ndarray] | None = None):
         import queue
         import threading
         workers = max(1, int(workers))
         self.qs: list["queue.Queue"] = [queue.Queue(maxsize=max(1, depth // workers)) for _ in range(workers)]
         self._threads = []
         for w in range(workers):
+            s = PatchSampler(images, targets, cfg, seed=seed * 1000 + w, focus=focus)
             th = threading.Thread(target=self._work, args=(s, self.qs[w]), daemon=True)
             th.start()
             self._threads.append(th)

cellquant/infer.py CHANGED Viewed

@@ -5,7 +5,7 @@ import numpy as np
 import torch
 from .annotations import to_rgb_uint8
-from .model import CellUNet, load_checkpoint, pad_to_multiple
 from .postprocess import PostConfig, Result, analyze, logits_to_probs
@@ -49,16 +49,25 @@ class Predictor:
         assert tta in (1, 4, 8)
         self.tta = tta
         self.tile, self.overlap, self.max_whole = tile, overlap, max_whole
     @torch.no_grad()
     def _probs_tensor(self, x: torch.Tensor) -> torch.Tensor:
-        """x: (1, 3, H, W) float in [0,1] on device, H, W multiples of 32 -> (8-ch probabilities) (1, 8, H, W)."""
         transforms = DIHEDRAL[:1] if self.tta == 1 else ([(0, False), (2, False), (0, True), (2, True)] if self.tta == 4 else DIHEDRAL)
         acc = None
         for k, f in transforms:
             out = self.model(_fwd_t(x, k, f))
             out = _inv_t(out, k, f)
-            p = torch.cat([out[:, 0:3].softmax(1), out[:, 3:4].sigmoid(), out[:, 4:5].sigmoid(), out[:, 5:8].softmax(1)], dim=1)
             acc = p if acc is None else acc + p
         return acc / len(transforms)
@@ -72,14 +81,17 @@ class Predictor:
             p = self._probs_tensor(xp.to(self.device))[0, :, :H, :W].cpu().numpy()
         else:
             p = self._tiled(x)[:, :H, :W]
-        return {"cell": p[0:3], "antibody": p[3], "cellnos": p[4], "nos": p[5:8]}
     def _tiled(self, x: torch.Tensor) -> np.ndarray:
         """Overlapping tiles blended with a linear ramp (no visible seams)."""
         _, _, H, W = x.shape
         t, o = self.tile, self.overlap
         step = t - o
-        acc = np.zeros((8, H, W), np.float32)
         wsum = np.zeros((1, H, W), np.float32)
         ramp = np.minimum(np.arange(t) + 1, np.arange(t)[::-1] + 1).astype(np.float32)
         ramp = np.minimum(ramp, o) / o

 import torch
 from .annotations import to_rgb_uint8
+from .model import DIST_CH, CellUNet, load_checkpoint, pad_to_multiple
 from .postprocess import PostConfig, Result, analyze, logits_to_probs
         assert tta in (1, 4, 8)
         self.tta = tta
         self.tile, self.overlap, self.max_whole = tile, overlap, max_whole
+        self.n_ch = int(self.model.head.out_channels)
+    @property
+    def has_dist(self) -> bool:
+        """True when the model predicts the distance channel (checkpoints trained after v1)."""
+        return self.n_ch > DIST_CH
     @torch.no_grad()
     def _probs_tensor(self, x: torch.Tensor) -> torch.Tensor:
+        """x: (1, 3, H, W) float in [0,1] on device, H, W multiples of 32 -> probabilities (1, 8 or 9, H, W)."""
         transforms = DIHEDRAL[:1] if self.tta == 1 else ([(0, False), (2, False), (0, True), (2, True)] if self.tta == 4 else DIHEDRAL)
         acc = None
         for k, f in transforms:
             out = self.model(_fwd_t(x, k, f))
             out = _inv_t(out, k, f)
+            parts = [out[:, 0:3].softmax(1), out[:, 3:4].sigmoid(), out[:, 4:5].sigmoid(), out[:, 5:8].softmax(1)]
+            if self.has_dist:
+                parts.append(out[:, DIST_CH:DIST_CH + 1].clamp(0.0, 1.0))
+            p = torch.cat(parts, dim=1)
             acc = p if acc is None else acc + p
         return acc / len(transforms)
             p = self._probs_tensor(xp.to(self.device))[0, :, :H, :W].cpu().numpy()
         else:
             p = self._tiled(x)[:, :H, :W]
+        probs = {"cell": p[0:3], "antibody": p[3], "cellnos": p[4], "nos": p[5:8]}
+        if self.has_dist:
+            probs["dist"] = p[DIST_CH]
+        return probs
     def _tiled(self, x: torch.Tensor) -> np.ndarray:
         """Overlapping tiles blended with a linear ramp (no visible seams)."""
         _, _, H, W = x.shape
         t, o = self.tile, self.overlap
         step = t - o
+        acc = np.zeros((self.n_ch, H, W), np.float32)
         wsum = np.zeros((1, H, W), np.float32)
         ramp = np.minimum(np.arange(t) + 1, np.arange(t)[::-1] + 1).astype(np.float32)
         ramp = np.minimum(ramp, o) / o

cellquant/losses.py CHANGED Viewed

@@ -4,7 +4,17 @@ from __future__ import annotations
 import torch
 import torch.nn.functional as F
-from .model import ANTIBODY_CH, CELLNOS_CH, CELL_SLICE, NOS_SLICE
 def soft_dice(probs: torch.Tensor, target_onehot: torch.Tensor, eps: float = 1.0) -> torch.Tensor:
@@ -15,19 +25,31 @@ def soft_dice(probs: torch.Tensor, target_onehot: torch.Tensor, eps: float = 1.0
     return (1 - (2 * inter + eps) / (denom + eps)).mean()
-def masked_bce(logit: torch.Tensor, target: torch.Tensor, mask: torch.Tensor) -> torch.Tensor:
     l = F.binary_cross_entropy_with_logits(logit, target, reduction="none")
-    return (l * mask).sum() / mask.sum().clamp_min(1.0)
-def multitask_loss(out: torch.Tensor, t: dict[str, torch.Tensor]) -> tuple[torch.Tensor, dict[str, float]]:
     dev = out.device
     cell_logits = out[:, CELL_SLICE]
     nos_logits = out[:, NOS_SLICE]
     cell3 = t["cell3"].to(dev)
     nos3 = t["nos3"].to(dev)
-    w_cell = torch.tensor([1.0, 1.0, 2.0], device=dev)
-    w_nos = torch.tensor([1.0, 4.0, 4.0], device=dev)
     ce_cell = F.cross_entropy(cell_logits, cell3, weight=w_cell)
     ce_nos = F.cross_entropy(nos_logits, nos3, weight=w_nos)
     p_cell = cell_logits.softmax(1)
@@ -37,12 +59,17 @@ def multitask_loss(out: torch.Tensor, t: dict[str, torch.Tensor]) -> tuple[torch
     dice_cell = soft_dice(p_cell[:, 1:], oh_cell[:, 1:])
     dice_nos = soft_dice(p_nos[:, 1:], oh_nos[:, 1:])
     mask = t["cellmask"].to(dev)
-    bce_ab = masked_bce(out[:, ANTIBODY_CH], t["antibody"].to(dev), mask)
     bce_cn = masked_bce(out[:, CELLNOS_CH], t["cellnos"].to(dev), mask)
     total = ce_cell + dice_cell + ce_nos + dice_nos + bce_ab + bce_cn
     parts = {
         "ce_cell": ce_cell.item(), "dice_cell": dice_cell.item(),
         "ce_nos": ce_nos.item(), "dice_nos": dice_nos.item(),
-        "bce_ab": bce_ab.item(), "bce_cn": bce_cn.item(), "total": total.item(),
     }
     return total, parts

 import torch
 import torch.nn.functional as F
+from .model import ANTIBODY_CH, CELLNOS_CH, CELL_SLICE, DIST_CH, NOS_SLICE
+# Class weights of the nucleus map (background / interior / border). The border ring
+# is 3 px wide (data.BORDER_WIDTH) and separates touching nuclei, so it gets the
+# largest weight.
+W_CELL = (1.0, 1.0, 3.0)
+W_NOS = (1.0, 4.0, 4.0)
+# Weight of the distance-regression term and of background pixels inside it
+# (nucleus pixels have weight 1; the watershed only ever looks inside nuclei).
+LAMBDA_DIST = 5.0
+DIST_BG_WEIGHT = 0.1
 def soft_dice(probs: torch.Tensor, target_onehot: torch.Tensor, eps: float = 1.0) -> torch.Tensor:
     return (1 - (2 * inter + eps) / (denom + eps)).mean()
+def masked_bce(logit: torch.Tensor, target: torch.Tensor, mask: torch.Tensor, neg_weight: float = 1.0) -> torch.Tensor:
+    """BCE averaged over ``mask`` pixels. ``neg_weight`` > 1 up-weights pixels whose
+    target is 0 (class balancing for a rare negative class); the result is normalised
+    by the summed weights so its magnitude does not depend on the weight."""
     l = F.binary_cross_entropy_with_logits(logit, target, reduction="none")
+    w = mask if neg_weight == 1.0 else mask * torch.where(target > 0.5, 1.0, float(neg_weight))
+    return (l * w).sum() / w.sum().clamp_min(1.0)
+def distance_loss(pred: torch.Tensor, target: torch.Tensor, cellmask: torch.Tensor) -> torch.Tensor:
+    """Weighted mean squared error between the predicted and the target distance map."""
+    w = cellmask + DIST_BG_WEIGHT * (1.0 - cellmask)
+    return ((pred - target) ** 2 * w).sum() / w.sum().clamp_min(1.0)
+def multitask_loss(out: torch.Tensor, t: dict[str, torch.Tensor], ab_neg_weight: float = 1.0) -> tuple[torch.Tensor, dict[str, float]]:
+    """``out``: (B, 8 or 9, H, W) network output; ``t``: targets from data.make_targets.
+    ``ab_neg_weight`` balances the antibody head towards antibody-negative pixels."""
     dev = out.device
     cell_logits = out[:, CELL_SLICE]
     nos_logits = out[:, NOS_SLICE]
     cell3 = t["cell3"].to(dev)
     nos3 = t["nos3"].to(dev)
+    w_cell = torch.tensor(W_CELL, device=dev)
+    w_nos = torch.tensor(W_NOS, device=dev)
     ce_cell = F.cross_entropy(cell_logits, cell3, weight=w_cell)
     ce_nos = F.cross_entropy(nos_logits, nos3, weight=w_nos)
     p_cell = cell_logits.softmax(1)
     dice_cell = soft_dice(p_cell[:, 1:], oh_cell[:, 1:])
     dice_nos = soft_dice(p_nos[:, 1:], oh_nos[:, 1:])
     mask = t["cellmask"].to(dev)
+    bce_ab = masked_bce(out[:, ANTIBODY_CH], t["antibody"].to(dev), mask, neg_weight=ab_neg_weight)
     bce_cn = masked_bce(out[:, CELLNOS_CH], t["cellnos"].to(dev), mask)
     total = ce_cell + dice_cell + ce_nos + dice_nos + bce_ab + bce_cn
     parts = {
         "ce_cell": ce_cell.item(), "dice_cell": dice_cell.item(),
         "ce_nos": ce_nos.item(), "dice_nos": dice_nos.item(),
+        "bce_ab": bce_ab.item(), "bce_cn": bce_cn.item(),
     }
+    if out.shape[1] > DIST_CH and "dist" in t:
+        mse_dist = distance_loss(out[:, DIST_CH], t["dist"].to(dev), mask)
+        total = total + LAMBDA_DIST * mse_dist
+        parts["mse_dist"] = mse_dist.item()
+    parts["total"] = total.item()
     return total, parts

cellquant/model.py CHANGED Viewed

@@ -5,6 +5,13 @@ Output channels (all at full input resolution):
     3     antibody logit    : P(antibody positive)      (meaningful on nucleus pixels)
     4     nostril-attr logit: P(nucleus contains punctum) (meaningful on nucleus pixels)
     5..7  punctum classes   : background / in nucleus / external (softmax)
 """
 from __future__ import annotations
@@ -12,11 +19,13 @@ import torch
 import torch.nn as nn
 import torch.nn.functional as F
-N_OUT = 8
 CELL_SLICE = slice(0, 3)
 ANTIBODY_CH = 3
 CELLNOS_CH = 4
 NOS_SLICE = slice(5, 8)
 # ImageNet statistics; inputs are RGB in [0, 1].
 MEAN = (0.485, 0.456, 0.406)

     3     antibody logit    : P(antibody positive)      (meaningful on nucleus pixels)
     4     nostril-attr logit: P(nucleus contains punctum) (meaningful on nucleus pixels)
     5..7  punctum classes   : background / in nucleus / external (softmax)
+    8     distance          : normalised distance to the nucleus outline (regression,
+                              1 at the centre of each nucleus, 0 at its outline and
+                              outside); the watershed landscape in postprocess.py
+Checkpoints written before channel 8 existed carry ``n_out = 8``; load_checkpoint
+rebuilds them without the distance head and post-processing falls back to the
+interior map.
 """
 from __future__ import annotations
 import torch.nn as nn
 import torch.nn.functional as F
+N_OUT = 9
 CELL_SLICE = slice(0, 3)
 ANTIBODY_CH = 3
 CELLNOS_CH = 4
 NOS_SLICE = slice(5, 8)
+DIST_CH = 8
+N_OUT_LEGACY = 8  # checkpoints without the distance channel
 # ImageNet statistics; inputs are RGB in [0, 1].
 MEAN = (0.485, 0.456, 0.406)

cellquant/postprocess.py CHANGED Viewed

@@ -8,13 +8,14 @@ from scipy import ndimage as ndi
 from skimage.segmentation import find_boundaries, watershed
 from .labels import ALL_LABELS, LABEL_COLORS, cell_label, nostril_label
-from .model import ANTIBODY_CH, CELLNOS_CH, CELL_SLICE, NOS_SLICE
 @dataclass
 class PostConfig:
     cell_fg_thr: float = 0.5       # P(interior) + P(border) > thr  -> nucleus foreground
-    marker_thr: float = 0.5        # P(interior) > thr               -> watershed markers
     min_marker_area: int = 12      # px
     min_cell_area: int = 50        # px (smallest traced nucleus in the training set is 35 px)
     nos_fg_thr: float = 0.5        # 1 - P(background) > thr         -> punctum foreground
@@ -27,33 +28,43 @@ class PostConfig:
 def logits_to_probs(logits: np.ndarray) -> dict[str, np.ndarray]:
-    """logits: (8, H, W) float -> dict of probabilities."""
     def softmax(z):
         z = z - z.max(axis=0, keepdims=True)
         e = np.exp(z)
         return e / e.sum(axis=0, keepdims=True)
     def sigmoid(z):
         return 1.0 / (1.0 + np.exp(-z))
-    return {
         "cell": softmax(logits[CELL_SLICE].astype(np.float32)),
         "antibody": sigmoid(logits[ANTIBODY_CH].astype(np.float32)),
         "cellnos": sigmoid(logits[CELLNOS_CH].astype(np.float32)),
         "nos": softmax(logits[NOS_SLICE].astype(np.float32)),
     }
-def segment_cells(p_cell: np.ndarray, cfg: PostConfig) -> np.ndarray:
-    """p_cell: (3, H, W) softmax (bg, interior, border) -> int32 instance map."""
     p_int, p_bor = p_cell[1], p_cell[2]
     fg = (p_int + p_bor) > cfg.cell_fg_thr
-    markers, n = ndi.label(p_int > cfg.marker_thr)
     if n:
         sizes = np.bincount(markers.ravel())
         small = sizes < cfg.min_marker_area
         small[0] = False
         markers[small[markers]] = 0
         markers, n = ndi.label(markers > 0)
-    inst = watershed(-p_int, markers, mask=fg) if n else np.zeros(fg.shape, np.int32)
     inst = inst.astype(np.int32)
     # foreground components that received no marker become instances of their own
     orphan, m = ndi.label(fg & (inst == 0))
@@ -129,7 +140,7 @@ class Result:
 def analyze(probs: dict[str, np.ndarray], cfg: PostConfig | None = None) -> Result:
     cfg = cfg or PostConfig()
-    cell_inst = segment_cells(probs["cell"], cfg)
     nos_inst = segment_nostrils(probs["nos"], cfg)
     n_cells = int(cell_inst.max())
     n_nos = int(nos_inst.max())

 from skimage.segmentation import find_boundaries, watershed
 from .labels import ALL_LABELS, LABEL_COLORS, cell_label, nostril_label
+from .model import ANTIBODY_CH, CELLNOS_CH, CELL_SLICE, DIST_CH, NOS_SLICE
 @dataclass
 class PostConfig:
     cell_fg_thr: float = 0.5       # P(interior) + P(border) > thr  -> nucleus foreground
+    marker_thr: float = 0.5        # predicted distance (or, for models without the distance channel,
+                                   # P(interior)) > thr               -> watershed markers
     min_marker_area: int = 12      # px
     min_cell_area: int = 50        # px (smallest traced nucleus in the training set is 35 px)
     nos_fg_thr: float = 0.5        # 1 - P(background) > thr         -> punctum foreground
 def logits_to_probs(logits: np.ndarray) -> dict[str, np.ndarray]:
+    """logits: (8 or 9, H, W) float -> dict of probabilities; the optional 9th channel
+    (normalised distance to the nucleus outline) is clipped to [0, 1] as "dist"."""
     def softmax(z):
         z = z - z.max(axis=0, keepdims=True)
         e = np.exp(z)
         return e / e.sum(axis=0, keepdims=True)
     def sigmoid(z):
         return 1.0 / (1.0 + np.exp(-z))
+    out = {
         "cell": softmax(logits[CELL_SLICE].astype(np.float32)),
         "antibody": sigmoid(logits[ANTIBODY_CH].astype(np.float32)),
         "cellnos": sigmoid(logits[CELLNOS_CH].astype(np.float32)),
         "nos": softmax(logits[NOS_SLICE].astype(np.float32)),
     }
+    if logits.shape[0] > DIST_CH:
+        out["dist"] = np.clip(logits[DIST_CH].astype(np.float32), 0.0, 1.0)
+    return out
+def segment_cells(p_cell: np.ndarray, cfg: PostConfig, dist: np.ndarray | None = None) -> np.ndarray:
+    """p_cell: (3, H, W) softmax (bg, interior, border) -> int32 instance map.
+    Foreground always comes from the class map. Markers and the watershed
+    landscape come from ``dist`` (the predicted normalised distance to the
+    outline, which has one peak per nucleus and a valley between touching ones)
+    when the model provides it, otherwise from P(interior)."""
     p_int, p_bor = p_cell[1], p_cell[2]
     fg = (p_int + p_bor) > cfg.cell_fg_thr
+    land = p_int if dist is None else dist
+    markers, n = ndi.label((land > cfg.marker_thr) & fg)
     if n:
         sizes = np.bincount(markers.ravel())
         small = sizes < cfg.min_marker_area
         small[0] = False
         markers[small[markers]] = 0
         markers, n = ndi.label(markers > 0)
+    inst = watershed(-land, markers, mask=fg) if n else np.zeros(fg.shape, np.int32)
     inst = inst.astype(np.int32)
     # foreground components that received no marker become instances of their own
     orphan, m = ndi.label(fg & (inst == 0))
 def analyze(probs: dict[str, np.ndarray], cfg: PostConfig | None = None) -> Result:
     cfg = cfg or PostConfig()
+    cell_inst = segment_cells(probs["cell"], cfg, probs.get("dist"))
     nos_inst = segment_nostrils(probs["nos"], cfg)
     n_cells = int(cell_inst.max())
     n_nos = int(nos_inst.max())

reports/SUMMARY.md CHANGED Viewed

@@ -3,70 +3,103 @@
 Data: 9 annotated 1024×1024 images, 1569 nucleus outlines and 1057 punctum outlines.
 Protocol: leave-one-image-out cross-validation (one model per image, 3000 iterations each, held-out image never seen in training), then a final model trained on all images with the same recipe. Each held-out image was tested 23 times: 3 identical runs (determinism), no-TTA / 4× / 8× test-time augmentation, 8 rotations and flips, 6 photometric perturbations, and 6 post-processing rule variants. The only choice made on held-out data is the rule that gives a nucleus the Nostril attribute: of three candidate rules, 'contains an in-nucleus punctum' (the annotation protocol) scored best and is the default; see the rule-variant table in cv_report.md.
 ## Headline numbers (micro-averaged over all held-out images, no TTA)
 | Measure | Precision | Recall | F1 |
 |---|---:|---:|---:|
-| Nucleus detection, IoU ≥ 0.5 | 77.5% | 78.4% | 77.9% |
-| Nucleus detection, IoU ≥ 0.3 | 85.6% | 86.7% | 86.2% |
-| Nucleus detection, centroid inside GT nucleus | 87.0% | 88.0% | 87.5% |
-| Punctum detection (centroid ≤ 6 px) | 77.3% | 82.8% | 79.9% |
-| Nucleus detection, IoU ≥ 0.5, 8× TTA | 78.1% | 78.9% | 78.5% |
 Per-class F1 (a true positive needs a matched object **and** the right label):
 | Label | GT | Pred | F1 |
 |---|---:|---:|---:|
-| Cell, Antibody + | 1061 | 1179 | 72.5% |
-| Cell, Antibody - | 105 | 39 | 31.9% |
-| Cell, Nostril, Antibody + | 398 | 370 | 71.6% |
 | Cell, Nostril, Antibody - | 5 | 0 | 0.0% |
-| Nostril in Nucleus | 525 | 529 | 78.9% |
-| Nostril External of Nucleus | 532 | 603 | 70.3% |
 ## Per held-out image
 | Image | Nuclei F1 | Antibody acc | Nostril-attr acc | Puncta F1 | Nuclei F1 over 14 perturbed runs (mean ± sd) | Deterministic |
 |---|---:|---:|---:|---:|---|---|
-| C100 | 81.2% | 99.3% | 98.6% | 74.4% | 82.5 ± 1.0 | yes |
-| C102 | 90.6% | 100.0% | 90.8% | 76.3% | 89.9 ± 1.8 | yes |
-| C110 | 78.5% | 100.0% | 91.9% | 84.3% | 77.5 ± 1.1 | yes |
-| C112 | 75.0% | 96.6% | 97.4% | 75.8% | 73.6 ± 1.5 | yes |
-| F100 | 75.1% | 97.8% | 97.1% | 85.2% | 72.2 ± 1.3 | yes |
-| F110 | 72.0% | 96.0% | 94.9% | 78.7% | 71.7 ± 1.9 | yes |
-| F113 | 65.4% | 100.0% | 97.1% | 85.2% | 65.6 ± 2.1 | yes |
-| F114 | 78.8% | 93.3% | 94.4% | 67.5% | 77.8 ± 1.7 | yes |
-| F99 | 84.4% | 77.5% | 99.5% | 71.4% | 85.3 ± 1.2 | yes |
 ## Final model (trained on all images) – fit on its own training images, NOT a generalisation estimate
 | Image | Nuclei F1 | Attribute acc | Puncta F1 |
 |---|---:|---:|---:|
-| C100 | 86.7% | 98.7% | 82.1% |
-| C102 | 93.3% | 94.4% | 84.6% |
-| C110 | 88.6% | 95.7% | 85.0% |
-| C112 | 83.8% | 99.2% | 86.2% |
-| F100 | 87.2% | 96.2% | 90.7% |
-| F110 | 91.7% | 94.8% | 87.2% |
-| F113 | 83.7% | 97.7% | 87.1% |
-| F114 | 88.3% | 93.4% | 83.1% |
-| F99 | 89.7% | 91.2% | 75.7% |
-## Antibody − threshold (held-out matched nuclei: 1230, of which 87 negative)
-Default threshold 0.5. Best Antibody − F1 in the sweep: 52.3% at threshold 0.85 (precision 62.9%, recall 44.8%); the full curve is in cv_report.md. The default was kept because the class is dominated by a single image (F99 holds 62 of the 102 negatives).
 ## What limits the scores
-* **Touching nuclei are sometimes merged** into one outline (worst on F113); this is the main source of IoU-0.5 misses. The marker/threshold sweep (`runs/cv/postprocess_sweep.json`) is flat, so it is a network-border issue, not a post-processing one.
-* **Annotation completeness**: most unmatched predictions overlap an annotated nucleus (outline disagreement) and the rest are dimmer, smaller objects the annotator skipped; many faint red dots were likewise not annotated, which caps external-punctum precision.
-* **Antibody − is rare** (102 of 1569 nuclei, 62 in one image), and **Cell, Nostril, Antibody −** has five examples; their scores are not reliable.
 * Scores are from nine images of one experiment (35-day spinal cord, Iba1/TMEM119). New staining batches or magnifications need a check against a few annotated images.
 ## Reproduce
 ```bash
-python evaluate.py cv --data-dir <folder with TIFs and CVAT zips> --runs-dir runs/cv --iters 3000
 python evaluate.py final --data-dir <folder> --weights weights/cellquant.pt
-python scripts/build_summary.py
 ```

 Data: 9 annotated 1024×1024 images, 1569 nucleus outlines and 1057 punctum outlines.
 Protocol: leave-one-image-out cross-validation (one model per image, 3000 iterations each, held-out image never seen in training), then a final model trained on all images with the same recipe. Each held-out image was tested 23 times: 3 identical runs (determinism), no-TTA / 4× / 8× test-time augmentation, 8 rotations and flips, 6 photometric perturbations, and 6 post-processing rule variants. The only choice made on held-out data is the rule that gives a nucleus the Nostril attribute: of three candidate rules, 'contains an in-nucleus punctum' (the annotation protocol) scored best and is the default; see the rule-variant table in cv_report.md.
+Recipe (v2, shipped): ResNet34-UNet with a full-resolution stem; targets are a 3-px inner border ring (class weight 3), a per-nucleus normalised distance-to-outline map that the watershed uses as its landscape (markers = distance > 0.5), an antibody head whose loss up-weights antibody-negative pixels by sqrt(positive / negative pixels) (4.5 on the full set), and 25 % of training crops centred on an antibody-negative nucleus. v1 (reports/cv/) used a 2-px ring, the interior map as watershed landscape and an unweighted antibody loss; the change table below compares the two under the same protocol.
 ## Headline numbers (micro-averaged over all held-out images, no TTA)
 | Measure | Precision | Recall | F1 |
 |---|---:|---:|---:|
+| Nucleus detection, IoU ≥ 0.5 | 75.8% | 85.5% | 80.4% |
+| Nucleus detection, IoU ≥ 0.3 | 82.7% | 93.3% | 87.7% |
+| Nucleus detection, centroid inside GT nucleus | 83.7% | 94.5% | 88.8% |
+| Punctum detection (centroid ≤ 6 px) | 76.4% | 84.8% | 80.4% |
+| Nucleus detection, IoU ≥ 0.5, 8× TTA | 75.9% | 85.7% | 80.5% |
 Per-class F1 (a true positive needs a matched object **and** the right label):
 | Label | GT | Pred | F1 |
 |---|---:|---:|---:|
+| Cell, Antibody + | 1061 | 1292 | 74.4% |
+| Cell, Antibody - | 105 | 91 | 37.8% |
+| Cell, Nostril, Antibody + | 398 | 388 | 72.8% |
 | Cell, Nostril, Antibody - | 5 | 0 | 0.0% |
+| Nostril in Nucleus | 525 | 532 | 78.7% |
+| Nostril External of Nucleus | 532 | 641 | 71.8% |
 ## Per held-out image
 | Image | Nuclei F1 | Antibody acc | Nostril-attr acc | Puncta F1 | Nuclei F1 over 14 perturbed runs (mean ± sd) | Deterministic |
 |---|---:|---:|---:|---:|---|---|
+| C100 | 84.4% | 96.9% | 98.1% | 71.4% | 85.5 ± 1.5 | yes |
+| C102 | 86.4% | 100.0% | 89.5% | 78.9% | 86.6 ± 1.2 | yes |
+| C110 | 77.4% | 98.7% | 94.2% | 83.8% | 78.4 ± 1.5 | yes |
+| C112 | 71.6% | 95.0% | 97.5% | 76.9% | 71.0 ± 1.4 | yes |
+| F100 | 82.9% | 98.1% | 94.4% | 85.7% | 78.8 ± 2.8 | yes |
+| F110 | 77.0% | 97.2% | 90.8% | 82.9% | 71.7 ± 2.5 | yes |
+| F113 | 69.9% | 100.0% | 97.4% | 83.3% | 70.6 ± 1.5 | yes |
+| F114 | 82.4% | 90.1% | 94.1% | 63.5% | 81.8 ± 0.8 | yes |
+| F99 | 87.5% | 79.3% | 99.6% | 74.3% | 87.8 ± 0.7 | yes |
 ## Final model (trained on all images) – fit on its own training images, NOT a generalisation estimate
 | Image | Nuclei F1 | Attribute acc | Puncta F1 |
 |---|---:|---:|---:|
+| C100 | 93.0% | 98.3% | 80.0% |
+| C102 | 91.5% | 94.5% | 78.8% |
+| C110 | 87.0% | 96.4% | 83.9% |
+| C112 | 80.0% | 99.2% | 82.0% |
+| F100 | 88.7% | 94.8% | 90.3% |
+| F110 | 89.7% | 97.5% | 88.6% |
+| F113 | 84.8% | 97.9% | 92.1% |
+| F114 | 90.8% | 95.1% | 82.5% |
+| F99 | 91.5% | 97.1% | 90.9% |
+## Change from v1 (same protocol, same nine held-out images; `reports/cv_v2/comparison.md` has the full table)
+| Measure | v1 | v2 | Δ (points) |
+|---|---:|---:|---:|
+| Nucleus F1, IoU ≥ 0.5 | 77.9% | 80.4% | +2.4 |
+| Nucleus precision, IoU ≥ 0.5 | 77.5% | 75.8% | -1.7 |
+| Nucleus recall, IoU ≥ 0.5 | 78.4% | 85.5% | +7.1 |
+| Nucleus F1, centroid rule | 87.5% | 88.8% | +1.3 |
+| Nucleus F1, IoU ≥ 0.5, 8× TTA | 78.5% | 80.5% | +2.0 |
+| Puncta F1 | 79.9% | 80.4% | +0.4 |
+| F1 [Cell, Antibody +] | 72.5% | 74.4% | +1.9 |
+| F1 [Cell, Antibody -] | 31.9% | 37.8% | +5.8 |
+| F1 [Cell, Nostril, Antibody +] | 71.6% | 72.8% | +1.2 |
+| F1 [Cell, Nostril, Antibody -] | 0.0% | 0.0% | +0.0 |
+| F1 [Nostril in Nucleus] | 78.9% | 78.7% | -0.2 |
+| F1 [Nostril External of Nucleus] | 70.3% | 71.8% | +1.5 |
+| Attribute accuracy, both (each model's own matched nuclei) | 90.2% | 89.3% | -1.0 |
+| Punctum type accuracy (matched puncta) | 93.1% | 93.4% | +0.3 |
+| Nuclei F1, C100 held out | 81.2% | 84.4% | +3.2 |
+| Nuclei F1, C102 held out | 90.6% | 86.4% | -4.2 |
+| Nuclei F1, C110 held out | 78.5% | 77.4% | -1.1 |
+| Nuclei F1, C112 held out | 75.0% | 71.6% | -3.4 |
+| Nuclei F1, F100 held out | 75.1% | 82.9% | +7.8 |
+| Nuclei F1, F110 held out | 72.0% | 77.0% | +5.0 |
+| Nuclei F1, F113 held out | 65.4% | 69.9% | +4.5 |
+| Nuclei F1, F114 held out | 78.8% | 82.4% | +3.7 |
+| Nuclei F1, F99 held out | 84.4% | 87.5% | +3.2 |
+v2 misses far fewer nuclei (339 → 227 unmatched annotated nuclei, mostly touching nuclei that v1 merged) at the cost of more unmatched predictions (358 → 429): it predicts 1771 nuclei for 1569 annotated ones (v1: 1588). The extra predictions are dim objects the annotator skipped and outline/split disagreements in roughly the proportion shown in the FP table of cv_report.md. On the nuclei matched by both models the attribute accuracy differs by less than one point (see comparison.md), which is within the sampling error of ~1200 nuclei; the antibody-negative class gains recall (26 % → 38 % on matched nuclei) and loses precision (74 % → 59 %).
+## Antibody − threshold (held-out matched nuclei: 1342, of which 98 negative)
+Default threshold 0.5. Best Antibody − F1 in the sweep: 50.8% at threshold 0.90 (precision 51.6%, recall 50.0%); the full curve is in cv_report.md. The default was kept because the class is dominated by a single image (F99 holds 62 of the 110 negatives).
 ## What limits the scores
+* **Touching nuclei** are still the main source of IoU-0.5 misses (worst on F113, the densest image). The distance-map watershed of v2 recovers many clusters that v1 merged (v1 marker/threshold sweeps in `reports/cv/postprocess_sweep.json` had shown post-processing alone could not), but some clusters are still merged and a few nuclei are now split in two.
+* **Annotation completeness**: most unmatched predictions overlap an annotated nucleus (outline disagreement) and the rest are dimmer, smaller objects the annotator skipped; many faint red dots were likewise not annotated, which caps external-punctum precision. v2 detects more of these dim objects than v1, so it over-counts nuclei by about 13 % on the annotated images.
+* **Antibody − is rare** (110 of 1569 nuclei, 62 in one image), and **Cell, Nostril, Antibody −** has five examples; their scores are not reliable. The class-balanced loss of v2 trades antibody-negative precision for recall.
 * Scores are from nine images of one experiment (35-day spinal cord, Iba1/TMEM119). New staining batches or magnifications need a check against a few annotated images.
 ## Reproduce
 ```bash
+python evaluate.py cv --data-dir <folder with TIFs and CVAT zips> --runs-dir runs/cv_v2 --iters 3000
 python evaluate.py final --data-dir <folder> --weights weights/cellquant.pt
+python scripts/compare_cv.py reports/cv/cv_results.json runs/cv_v2/cv_results.json --out reports/cv_v2/comparison.md
+python scripts/build_summary.py --cv-dir runs/cv_v2
 ```

reports/cv_v2/comparison.md ADDED Viewed

	@@ -0,0 +1,94 @@

+# Cross-validation comparison: v1 vs v2
+`reports/cv/cv_results.json` vs `runs/cv_v2/cv_results.json`; micro-averaged over the held-out images (base run, no TTA unless stated).
+| Measure | v1 | v2 | Δ (points) |
+|---|---:|---:|---:|
+| Nucleus F1, IoU ≥ 0.5 | 77.9% | 80.4% | +2.4 |
+| Nucleus precision, IoU ≥ 0.5 | 77.5% | 75.8% | -1.7 |
+| Nucleus recall, IoU ≥ 0.5 | 78.4% | 85.5% | +7.1 |
+| Nucleus F1, IoU ≥ 0.3 | 86.2% | 87.7% | +1.5 |
+| Nucleus F1, centroid rule | 87.5% | 88.8% | +1.3 |
+| Nucleus F1, IoU ≥ 0.5, 8× TTA | 78.5% | 80.5% | +2.0 |
+| Puncta F1 | 79.9% | 80.4% | +0.4 |
+| Puncta F1, 8× TTA | 81.4% | 81.8% | +0.5 |
+| Attribute accuracy (both, matched nuclei) | 90.2% | 89.3% | -1.0 |
+| Punctum type accuracy (matched puncta) | 93.1% | 93.4% | +0.3 |
+| F1 [Cell, Antibody +] | 72.5% | 74.4% | +1.9 |
+| F1 [Cell, Antibody -] | 31.9% | 37.8% | +5.8 |
+| F1 [Cell, Nostril, Antibody +] | 71.6% | 72.8% | +1.2 |
+| F1 [Cell, Nostril, Antibody -] | 0.0% | 0.0% | +0.0 |
+| F1 [Nostril in Nucleus] | 78.9% | 78.7% | -0.2 |
+| F1 [Nostril External of Nucleus] | 70.3% | 71.8% | +1.5 |
+| Precision [Cell, Antibody -] | 59.0% | 40.7% | -18.3 |
+| Recall [Cell, Antibody -] | 21.9% | 35.2% | +13.3 |
+## Per held-out image
+| Image | Nuclei F1 v1 | Nuclei F1 v2 | Δ | Centroid F1 v1 | Centroid F1 v2 | Antibody acc v1 | Antibody acc v2 | Puncta F1 v1 | Puncta F1 v2 |
+|---|---:|---:|---:|---:|---:|---:|---:|---:|---:|
+| C100 | 81.2% | 84.4% | +3.2 | 89.5% | 91.3% | 99.3% | 96.9% | 74.4% | 71.4% |
+| C102 | 90.6% | 86.4% | -4.2 | 92.7% | 90.5% | 100.0% | 100.0% | 76.3% | 78.9% |
+| C110 | 78.5% | 77.4% | -1.1 | 88.1% | 88.3% | 100.0% | 98.7% | 84.3% | 83.8% |
+| C112 | 75.0% | 71.6% | -3.4 | 87.2% | 83.4% | 96.6% | 95.0% | 75.8% | 76.9% |
+| F100 | 75.1% | 82.9% | +7.8 | 84.9% | 89.6% | 97.8% | 98.1% | 85.2% | 85.7% |
+| F110 | 72.0% | 77.0% | +5.0 | 85.1% | 86.9% | 96.0% | 97.2% | 78.7% | 82.9% |
+| F113 | 65.4% | 69.9% | +4.5 | 83.3% | 86.1% | 100.0% | 100.0% | 85.2% | 83.3% |
+| F114 | 78.8% | 82.4% | +3.7 | 84.5% | 89.4% | 93.3% | 90.1% | 67.5% | 63.5% |
+| F99 | 84.4% | 87.5% | +3.2 | 92.3% | 91.7% | 77.5% | 79.3% | 71.4% | 74.3% |
+## Robustness (nuclei F1 over the 14 rotated / perturbed runs per image)
+| Image | v1 mean ± sd | v2 mean ± sd |
+|---|---|---|
+| C100 | 82.5 ± 1.0 | 85.5 ± 1.5 |
+| C102 | 89.9 ± 1.8 | 86.6 ± 1.2 |
+| C110 | 77.5 ± 1.1 | 78.4 ± 1.5 |
+| C112 | 73.6 ± 1.5 | 71.0 ± 1.4 |
+| F100 | 72.2 ± 1.3 | 78.8 ± 2.8 |
+| F110 | 71.7 ± 1.9 | 71.7 ± 2.5 |
+| F113 | 65.6 ± 2.1 | 70.6 ± 1.5 |
+| F114 | 77.8 ± 1.7 | 81.8 ± 0.8 |
+| F99 | 85.3 ± 1.2 | 87.8 ± 0.7 |
+## Antibody-negative class
+| | v1 | v2 |
+|---|---:|---:|
+| Cell, Antibody -: GT / predicted / TP | 105 / 39 / 23 | 105 / 91 / 37 |
+| Cell, Nostril, Antibody -: GT / predicted / TP | 5 / 0 / 0 | 5 / 0 / 0 |
+| Antibody − F1 among matched nuclei at the default threshold 0.5 | 39.0% (P 74.2%, R 26.4%) | 46.0% (P 58.7%, R 37.8%) |
+| Best threshold in the sweep (not used) | 52.3% at 0.85 | 50.8% at 0.90 |
+## Attribute accuracy on the 1194 held-out nuclei matched by both v1 and v2 (IoU >= 0.5, base run)
+| | v1 | v2 |
+|---|---:|---:|
+| Antibody +/− correct | 94.1% | 93.8% |
+| Nostril attribute correct | 96.1% | 95.7% |
+| Both correct | 90.2% | 89.5% |
+| Antibody − on this set (v1): TP / FP / FN, precision, recall, F1 | 23 / 8 / 63, 74.2%, 26.7%, 39.3% | |
+| Antibody − on this set (v2): TP / FP / FN, precision, recall, F1 | 34 / 22 / 52, 60.7%, 39.5%, 47.9% | |
+| Image | matched by both | v1 antibody acc | v2 antibody acc | v1 Nostril acc | v2 Nostril acc |
+|---|---:|---:|---:|---:|---:|
+| C100 | 144 | 99.3% | 97.2% | 98.6% | 97.9% |
+| C102 | 83 | 100.0% | 100.0% | 90.4% | 89.2% |
+| C110 | 144 | 100.0% | 98.6% | 91.7% | 93.8% |
+| C112 | 113 | 96.5% | 94.7% | 98.2% | 97.3% |
+| F100 | 133 | 98.5% | 97.7% | 97.0% | 95.5% |
+| F110 | 94 | 95.7% | 98.9% | 94.7% | 91.5% |
+| F113 | 99 | 100.0% | 100.0% | 97.0% | 97.0% |
+| F114 | 176 | 93.2% | 91.5% | 94.9% | 94.9% |
+| F99 | 208 | 76.9% | 79.3% | 99.5% | 99.5% |
+## Counts summed over the nine held-out images (and summed absolute per-image count error)
+| Label | GT | v1 | v2 | abs error v1 | abs error v2 |
+|---|---:|---:|---:|---:|---:|
+| Cell, Antibody + | 1061 | 1179 | 1292 | 180 | 239 |
+| Cell, Antibody - | 105 | 39 | 91 | 74 | 62 |
+| Cell, Nostril, Antibody + | 398 | 370 | 388 | 56 | 60 |
+| Cell, Nostril, Antibody - | 5 | 0 | 0 | 5 | 5 |
+| Nostril in Nucleus | 525 | 529 | 532 | 60 | 65 |
+| Nostril External of Nucleus | 532 | 603 | 641 | 187 | 205 |

reports/cv_v2/cv_report.md ADDED Viewed

	@@ -0,0 +1,105 @@

+# Leave-one-image-out cross-validation
+9 folds over 9 annotated images (1569 nucleus and 1057 punctum outlines); each image scored by a model whose checkpoint records that image as held out. Iterations per fold: 3000.
+## Micro-averaged per-class scores (base run, no TTA)
+| Label | GT | Pred | TP | FP | FN | Precision | Recall | F1 |
+|---|---:|---:|---:|---:|---:|---:|---:|---:|
+| Cell, Antibody + | 1061 | 1292 | 875 | 417 | 186 |  67.7% |  82.5% |  74.4% |
+| Cell, Antibody - | 105 | 91 | 37 | 54 | 68 |  40.7% |  35.2% |  37.8% |
+| Cell, Nostril, Antibody + | 398 | 388 | 286 | 102 | 112 |  73.7% |  71.9% |  72.8% |
+| Cell, Nostril, Antibody - | 5 | 0 | 0 | 0 | 5 |   0.0% |   0.0% |   0.0% |
+| Nostril in Nucleus | 525 | 532 | 416 | 116 | 109 |  78.2% |  79.2% |  78.7% |
+| Nostril External of Nucleus | 532 | 641 | 421 | 220 | 111 |  65.7% |  79.1% |  71.8% |
+Class-agnostic detection (IoU >= 0.5 for nuclei; centroid distance for puncta):
+- base: nuclei P  75.8% R  85.5% F1  80.4% (IoU≥0.5); F1  87.7% (IoU≥0.3); F1  88.8% (centroid-in-nucleus) | puncta P  76.4% R  84.8% F1  80.4%
+- tta4: nuclei P  75.8% R  85.6% F1  80.4% (IoU≥0.5); F1  87.5% (IoU≥0.3); F1  89.0% (centroid-in-nucleus) | puncta P  77.8% R  85.7% F1  81.6%
+- tta8: nuclei P  75.9% R  85.7% F1  80.5% (IoU≥0.5); F1  87.7% (IoU≥0.3); F1  89.4% (centroid-in-nucleus) | puncta P  78.5% R  85.4% F1  81.8%
+## Per-image results (each image held out of training)
+| Image | Nuclei F1 (IoU≥0.5) | F1 (IoU≥0.3) | F1 (centroid) | mean IoU | Antibody acc | Nostril-attr acc | Puncta F1 | Puncta type acc | Deterministic |
+|---|---:|---:|---:|---:|---:|---:|---:|---:|---|
+| C100 |  84.4% |  91.3% |  91.3% | 0.784 |  96.9% |  98.1% |  71.4% | 100.0% | yes |
+| C102 |  86.4% |  90.5% |  90.5% | 0.811 | 100.0% |  89.5% |  78.9% |  90.8% | yes |
+| C110 |  77.4% |  85.4% |  88.3% | 0.747 |  98.7% |  94.2% |  83.8% |  93.9% | yes |
+| C112 |  71.6% |  81.1% |  83.4% | 0.723 |  95.0% |  97.5% |  76.9% |  92.0% | yes |
+| F100 |  82.9% |  89.6% |  89.6% | 0.742 |  98.1% |  94.4% |  85.7% |  97.7% | yes |
+| F110 |  77.0% |  82.0% |  86.9% | 0.776 |  97.2% |  90.8% |  82.9% |  90.2% | yes |
+| F113 |  69.9% |  84.9% |  86.1% | 0.717 | 100.0% |  97.4% |  83.3% |  88.0% | yes |
+| F114 |  82.4% |  89.4% |  89.4% | 0.794 |  90.1% |  94.1% |  63.5% |  95.0% | yes |
+| F99 |  87.5% |  91.7% |  91.7% | 0.791 |  79.3% |  99.6% |  74.3% |  92.3% | yes |
+## Repeated tests per image (8 rotations/flips + 6 photometric perturbations = 14 runs)
+| Image | Nuclei F1 mean ± sd [min, max] | Attr acc mean ± sd | Puncta F1 mean ± sd [min, max] |
+|---|---|---|---|
+| C100 | 85.5 ± 1.5 [83.2, 89.6] | 96.7 ± 1.1 | 72.1 ± 2.7 [68.2, 78.0] |
+| C102 | 86.6 ± 1.2 [84.6, 88.6] | 90.3 ± 2.1 | 78.8 ± 1.7 [75.7, 81.5] |
+| C110 | 78.4 ± 1.5 [75.5, 80.7] | 92.1 ± 1.1 | 84.2 ± 0.7 [82.7, 85.4] |
+| C112 | 71.0 ± 1.4 [68.1, 73.5] | 93.9 ± 1.6 | 75.3 ± 1.7 [72.5, 78.1] |
+| F100 | 78.8 ± 2.8 [72.8, 82.9] | 92.3 ± 2.6 | 84.5 ± 2.1 [78.7, 87.1] |
+| F110 | 71.7 ± 2.5 [66.7, 77.0] | 89.9 ± 2.4 | 81.8 ± 3.9 [73.6, 84.7] |
+| F113 | 70.6 ± 1.5 [67.7, 73.4] | 95.6 ± 1.6 | 81.7 ± 3.9 [72.7, 87.1] |
+| F114 | 81.8 ± 0.8 [80.2, 82.8] | 86.3 ± 1.6 | 66.4 ± 3.1 [62.6, 72.8] |
+| F99 | 87.8 ± 0.7 [86.3, 89.2] | 77.4 ± 1.7 | 70.8 ± 11.0 [53.3, 85.0] |
+## Why nuclei F1 is below 100 %: unmatched predictions and misses (base run)
+| Image | FP total | FP overlapping a GT nucleus | FP with no GT overlap | FP / matched median blue | FP / matched median area | FN total | FN covered by a prediction |
+|---|---:|---:|---:|---|---|---:|---:|
+| C100 | 41 | 31 | 10 | 44 / 79 | 247 / 624 | 18 | 16 |
+| C102 | 22 | 7 | 15 | 22 / 53 | 371 / 803 | 5 | 4 |
+| C110 | 63 | 38 | 25 | 17 / 42 | 371 / 628 | 28 | 23 |
+| C112 | 71 | 50 | 21 | 21 / 26 | 400 / 701 | 25 | 23 |
+| F100 | 40 | 22 | 18 | 15 / 34 | 574 / 790 | 26 | 24 |
+| F110 | 46 | 39 | 7 | 43 / 52 | 387 / 704 | 19 | 11 |
+| F113 | 53 | 32 | 21 | 9 / 22 | 594 / 757 | 47 | 45 |
+| F114 | 46 | 29 | 17 | 71 / 89 | 514 / 632 | 40 | 37 |
+| F99 | 47 | 38 | 9 | 44 / 53 | 446 / 596 | 19 | 15 |
+## Counts per image (ground truth / predicted, base run)
+| Image | Cell, Antibody + | Cell, Antibody - | Cell, Nostril, Antibody + | Cell, Nostril, Antibody - | Nostril in Nucleus | Nostril External of Nucleus |
+|---|---|---|---|---|---|---|
+| C100 | 165 / 177 | 0 / 13 | 13 / 11 | 0 / 0 | 14 / 15 | 6 / 7 |
+| C102 | 41 / 63 | 0 / 1 | 50 / 44 | 0 / 0 | 60 / 66 | 74 / 104 |
+| C110 | 61 / 91 | 1 / 3 | 122 / 125 | 0 / 0 | 169 / 176 | 207 / 267 |
+| C112 | 124 / 164 | 5 / 3 | 17 / 25 | 0 / 0 | 23 / 29 | 2 / 11 |
+| F100 | 87 / 105 | 6 / 14 | 93 / 81 | 0 / 0 | 118 / 119 | 44 / 27 |
+| F110 | 87 / 125 | 7 / 7 | 33 / 23 | 1 / 0 | 57 / 35 | 137 / 116 |
+| F113 | 138 / 148 | 3 / 2 | 22 / 19 | 0 / 0 | 26 / 21 | 8 / 5 |
+| F114 | 175 / 171 | 21 / 21 | 42 / 56 | 4 / 0 | 52 / 67 | 38 / 95 |
+| F99 | 183 / 248 | 62 / 27 | 6 / 4 | 0 / 0 | 6 / 4 | 16 / 9 |
+## Post-processing rule variants (micro F1 across held-out images; accuracies pooled over matched objects)
+| Variant | Cell, Antibody + | Cell, Antibody - | Cell, Nostril, Antibody + | Cell, Nostril, Antibody - | Nostril in Nucleus | Nostril External of Nucleus | attr acc (matched nuclei) | puncta type acc (matched puncta) |
+|---|---|---|---|---|---|---|---|---|
+| rule_nos=head_cell=head |  74.4% |  37.8% |  68.8% |   0.0% |  78.7% |  71.8% |  88.0% |  93.4% |
+| rule_nos=head_cell=contains |  74.4% |  37.8% |  72.8% |   0.0% |  78.7% |  71.8% |  89.3% |  93.4% |
+| rule_nos=head_cell=either |  74.5% |  37.8% |  69.1% |   0.0% |  78.7% |  71.8% |  88.2% |  93.4% |
+| rule_nos=geometric_cell=head |  74.4% |  37.8% |  68.8% |   0.0% |  79.0% |  71.6% |  88.0% |  93.4% |
+| rule_nos=geometric_cell=contains |  74.4% |  37.8% |  72.9% |   0.0% |  79.0% |  71.6% |  89.3% |  93.4% |
+| rule_nos=geometric_cell=either |  74.5% |  37.8% |  69.2% |   0.0% |  79.0% |  71.6% |  88.2% |  93.4% |
+## Antibody threshold sweep on the 1342 matched held-out nuclei (98 antibody-negative)
+| threshold | Antibody − precision | Antibody − recall | Antibody − F1 | overall accuracy |
+|---:|---:|---:|---:|---:|
+| 0.30 |  60.7% |  34.7% |  44.2% |  93.6% |
+| 0.35 |  61.4% |  35.7% |  45.2% |  93.7% |
+| 0.40 |  61.7% |  37.8% |  46.8% |  93.7% |
+| 0.45 |  59.7% |  37.8% |  46.2% |  93.6% |
+| 0.50 |  58.7% |  37.8% |  46.0% |  93.5% |
+| 0.55 |  58.5% |  38.8% |  46.6% |  93.5% |
+| 0.60 |  57.6% |  38.8% |  46.3% |  93.4% |
+| 0.65 |  56.5% |  39.8% |  46.7% |  93.4% |
+| 0.70 |  54.8% |  40.8% |  46.8% |  93.2% |
+| 0.75 |  51.9% |  41.8% |  46.3% |  92.9% |
+| 0.80 |  51.9% |  42.9% |  46.9% |  92.9% |
+| 0.85 |  52.8% |  48.0% |  50.3% |  93.1% |
+| 0.90 |  51.6% |  50.0% |  50.8% |  92.9% |

reports/cv_v2/cv_results.json ADDED Viewed

The diff for this file is too large to render. See raw diff

reports/cv_v2/run_all.sh ADDED Viewed

	@@ -0,0 +1,11 @@

+#!/bin/bash
+# Train the nine v2 folds (two at a time), then score every held-out image.
+cd "$(dirname "$0")/../.."
+Q=runs/cv_v2/train_queue.sh
+bash "$Q" C100 C102 C110 C112 F100 F110 F113 F114 F99 > runs/cv_v2/queue_A.log 2>&1 &
+bash "$Q" F99 F114 F113 F110 F100 C112 C110 C102 C100 > runs/cv_v2/queue_B.log 2>&1 &
+wait
+echo "=== training done $(date) ==="
+python3 -I evaluate.py cv --data-dir /Users/slf20757/Desktop/Peyton --runs-dir runs/cv_v2 --iters 3000 --device mps > runs/cv_v2/eval.log 2>&1
+echo "=== eval done $(date) exit $? ==="
+echo ALL_DONE

reports/cv_v2/train_queue.sh ADDED Viewed

	@@ -0,0 +1,11 @@

+#!/bin/bash
+# Lock-based fold queue for the v2 recipe: several instances may run concurrently with different orders.
+cd "$(dirname "$0")/../.."
+for K in "$@"; do
+  [ -f "runs/cv_v2/fold_$K.pt" ] && { echo "skip $K (done)"; continue; }
+  mkdir "runs/cv_v2/fold_$K.lock" 2>/dev/null || { echo "skip $K (claimed)"; continue; }
+  echo "=== fold $K start $(date) ==="
+  python3 -I train.py --data-dir /Users/slf20757/Desktop/Peyton --holdout "$K" --iters 3000 --val-every 500 --out "runs/cv_v2/fold_$K.pt" --device mps > "runs/cv_v2/train_$K.log" 2>&1 || echo "FOLD $K FAILED"
+  echo "=== fold $K done $(date) ==="
+done
+echo QUEUE_DONE

reports/final/final_fit.json CHANGED Viewed

@@ -2,56 +2,56 @@
  "C100": {
   "cells": {
    "detection": {
-    "tp": 153,
-    "fp": 22,
-    "fn": 25,
-    "precision": 0.8742857142857143,
-    "recall": 0.8595505617977528,
-    "f1": 0.8668555240793201,
-    "mean_iou_matched": 0.7974392864551416,
     "iou30": {
-     "tp": 165,
-     "fp": 10,
-     "fn": 13,
-     "precision": 0.9428571428571428,
-     "recall": 0.9269662921348315,
-     "f1": 0.934844192634561
     },
     "centroid": {
-     "tp": 165,
-     "fp": 10,
-     "fn": 13,
-     "precision": 0.9428571428571428,
-     "recall": 0.9269662921348315,
-     "f1": 0.934844192634561
     }
    },
    "attribute_accuracy": {
-    "antibody": 0.9934640522875817,
-    "nostril": 0.9934640522875817,
-    "both": 0.9869281045751634,
-    "n_matched": 153
    },
    "per_class": {
     "Cell, Antibody +": {
-     "tp": 139,
-     "fp": 22,
-     "fn": 26,
-     "precision": 0.8633540372670807,
-     "recall": 0.8424242424242424,
-     "f1": 0.852760736196319,
      "gt_count": 165,
-     "pred_count": 161
     },
     "Cell, Antibody -": {
      "tp": 0,
-     "fp": 2,
      "fn": 0,
      "precision": 0.0,
      "recall": 0.0,
      "f1": 0.0,
      "gt_count": 0,
-     "pred_count": 2
     },
     "Cell, Nostril, Antibody +": {
      "tp": 12,
@@ -76,8 +76,8 @@
    },
    "confusion": {
     "Cell, Antibody +": {
-     "Cell, Antibody +": 139,
-     "Cell, Antibody -": 1,
      "Cell, Nostril, Antibody +": 0,
      "Cell, Nostril, Antibody -": 0
     },
@@ -101,29 +101,29 @@
     }
    },
    "n_gt": 178,
-   "n_pred": 175
   },
   "nostrils": {
    "detection": {
     "tp": 16,
-    "fp": 3,
     "fn": 4,
-    "precision": 0.8421052631578947,
     "recall": 0.8,
-    "f1": 0.8205128205128205,
-    "mean_dist_matched": 1.157546253191996
    },
    "type_accuracy": 1.0,
    "per_class": {
     "Nostril in Nucleus": {
      "tp": 12,
-     "fp": 0,
      "fn": 2,
-     "precision": 1.0,
      "recall": 0.8571428571428571,
-     "f1": 0.923076923076923,
      "gt_count": 14,
-     "pred_count": 12
     },
     "Nostril External of Nucleus": {
      "tp": 4,
@@ -137,19 +137,19 @@
     }
    },
    "n_gt": 20,
-   "n_pred": 19
   },
   "counts": {
    "Cell, Antibody +": {
     "gt": 165,
-    "pred": 161,
-    "abs_err": 4,
-    "rel_err": 0.024242424242424242
    },
    "Cell, Antibody -": {
     "gt": 0,
-    "pred": 2,
-    "abs_err": 2,
     "rel_err": Infinity
    },
    "Cell, Nostril, Antibody +": {
@@ -166,9 +166,9 @@
    },
    "Nostril in Nucleus": {
     "gt": 14,
-    "pred": 12,
-    "abs_err": 2,
-    "rel_err": 0.14285714285714285
    },
    "Nostril External of Nucleus": {
     "gt": 6,
@@ -180,90 +180,90 @@
   "fp_analysis": null,
   "pairs": null,
   "summary": {
-   "cell_det_f1": 0.8668555240793201,
-   "cell_det_precision": 0.8742857142857143,
-   "cell_det_recall": 0.8595505617977528,
-   "cell_mean_iou": 0.7974392864551416,
-   "cell_det_f1_iou30": 0.934844192634561,
-   "cell_det_f1_centroid": 0.934844192634561,
-   "cell_attr_antibody": 0.9934640522875817,
-   "cell_attr_nostril": 0.9934640522875817,
-   "cell_attr_both": 0.9869281045751634,
-   "nos_det_f1": 0.8205128205128205,
-   "nos_det_precision": 0.8421052631578947,
    "nos_det_recall": 0.8,
    "nos_type_acc": 1.0,
-   "f1[Cell, Antibody +]": 0.852760736196319,
    "f1[Cell, Antibody -]": 0.0,
    "f1[Cell, Nostril, Antibody +]": 0.9600000000000001,
    "f1[Cell, Nostril, Antibody -]": 0.0,
-   "f1[Nostril in Nucleus]": 0.923076923076923,
    "f1[Nostril External of Nucleus]": 0.6153846153846153
   }
  },
  "C102": {
   "cells": {
    "detection": {
-    "tp": 90,
-    "fp": 12,
-    "fn": 1,
-    "precision": 0.8823529411764706,
-    "recall": 0.989010989010989,
-    "f1": 0.9326424870466321,
-    "mean_iou_matched": 0.8122454665114849,
     "iou30": {
-     "tp": 90,
-     "fp": 12,
-     "fn": 1,
-     "precision": 0.8823529411764706,
-     "recall": 0.989010989010989,
-     "f1": 0.9326424870466321
     },
     "centroid": {
-     "tp": 90,
-     "fp": 12,
-     "fn": 1,
-     "precision": 0.8823529411764706,
-     "recall": 0.989010989010989,
-     "f1": 0.9326424870466321
     }
    },
    "attribute_accuracy": {
     "antibody": 1.0,
-    "nostril": 0.9444444444444444,
-    "both": 0.9444444444444444,
-    "n_matched": 90
    },
    "per_class": {
     "Cell, Antibody +": {
      "tp": 41,
-     "fp": 15,
      "fn": 0,
-     "precision": 0.7321428571428571,
      "recall": 1.0,
-     "f1": 0.8453608247422679,
      "gt_count": 41,
-     "pred_count": 56
     },
     "Cell, Antibody -": {
      "tp": 0,
-     "fp": 0,
      "fn": 0,
      "precision": 0.0,
      "recall": 0.0,
      "f1": 0.0,
      "gt_count": 0,
-     "pred_count": 0
     },
     "Cell, Nostril, Antibody +": {
-     "tp": 44,
-     "fp": 2,
-     "fn": 6,
-     "precision": 0.9565217391304348,
-     "recall": 0.88,
-     "f1": 0.9166666666666666,
      "gt_count": 50,
-     "pred_count": 46
     },
     "Cell, Nostril, Antibody -": {
      "tp": 0,
@@ -292,7 +292,7 @@
     "Cell, Nostril, Antibody +": {
      "Cell, Antibody +": 5,
      "Cell, Antibody -": 0,
-     "Cell, Nostril, Antibody +": 44,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody -": {
@@ -303,62 +303,62 @@
     }
    },
    "n_gt": 91,
-   "n_pred": 102
   },
   "nostrils": {
    "detection": {
-    "tp": 121,
-    "fp": 31,
-    "fn": 13,
-    "precision": 0.7960526315789473,
-    "recall": 0.9029850746268657,
-    "f1": 0.8461538461538461,
-    "mean_dist_matched": 1.0261222516695998
-   },
-   "type_accuracy": 1.0,
    "per_class": {
     "Nostril in Nucleus": {
      "tp": 54,
-     "fp": 8,
      "fn": 6,
-     "precision": 0.8709677419354839,
      "recall": 0.9,
-     "f1": 0.8852459016393444,
      "gt_count": 60,
-     "pred_count": 62
     },
     "Nostril External of Nucleus": {
-     "tp": 67,
-     "fp": 23,
-     "fn": 7,
-     "precision": 0.7444444444444445,
-     "recall": 0.9054054054054054,
-     "f1": 0.8170731707317074,
      "gt_count": 74,
-     "pred_count": 90
     }
    },
    "n_gt": 134,
-   "n_pred": 152
   },
   "counts": {
    "Cell, Antibody +": {
     "gt": 41,
-    "pred": 56,
-    "abs_err": 15,
-    "rel_err": 0.36585365853658536
    },
    "Cell, Antibody -": {
     "gt": 0,
-    "pred": 0,
-    "abs_err": 0,
-    "rel_err": 0.0
    },
    "Cell, Nostril, Antibody +": {
     "gt": 50,
-    "pred": 46,
-    "abs_err": 4,
-    "rel_err": 0.08
    },
    "Cell, Nostril, Antibody -": {
     "gt": 0,
@@ -368,104 +368,104 @@
    },
    "Nostril in Nucleus": {
     "gt": 60,
-    "pred": 62,
-    "abs_err": 2,
-    "rel_err": 0.03333333333333333
    },
    "Nostril External of Nucleus": {
     "gt": 74,
-    "pred": 90,
-    "abs_err": 16,
-    "rel_err": 0.21621621621621623
    }
   },
   "fp_analysis": null,
   "pairs": null,
   "summary": {
-   "cell_det_f1": 0.9326424870466321,
-   "cell_det_precision": 0.8823529411764706,
-   "cell_det_recall": 0.989010989010989,
-   "cell_mean_iou": 0.8122454665114849,
-   "cell_det_f1_iou30": 0.9326424870466321,
-   "cell_det_f1_centroid": 0.9326424870466321,
    "cell_attr_antibody": 1.0,
-   "cell_attr_nostril": 0.9444444444444444,
-   "cell_attr_both": 0.9444444444444444,
-   "nos_det_f1": 0.8461538461538461,
-   "nos_det_precision": 0.7960526315789473,
-   "nos_det_recall": 0.9029850746268657,
-   "nos_type_acc": 1.0,
-   "f1[Cell, Antibody +]": 0.8453608247422679,
    "f1[Cell, Antibody -]": 0.0,
-   "f1[Cell, Nostril, Antibody +]": 0.9166666666666666,
    "f1[Cell, Nostril, Antibody -]": 0.0,
-   "f1[Nostril in Nucleus]": 0.8852459016393444,
-   "f1[Nostril External of Nucleus]": 0.8170731707317074
   }
  },
  "C110": {
   "cells": {
    "detection": {
-    "tp": 164,
-    "fp": 22,
-    "fn": 20,
-    "precision": 0.8817204301075269,
-    "recall": 0.8913043478260869,
-    "f1": 0.8864864864864864,
-    "mean_iou_matched": 0.800839882168248,
     "iou30": {
-     "tp": 173,
-     "fp": 13,
-     "fn": 11,
-     "precision": 0.9301075268817204,
-     "recall": 0.9402173913043478,
-     "f1": 0.9351351351351351
     },
     "centroid": {
-     "tp": 172,
-     "fp": 14,
-     "fn": 12,
-     "precision": 0.9247311827956989,
-     "recall": 0.9347826086956522,
-     "f1": 0.9297297297297297
     }
    },
    "attribute_accuracy": {
-    "antibody": 1.0,
-    "nostril": 0.9573170731707317,
-    "both": 0.9573170731707317,
-    "n_matched": 164
    },
    "per_class": {
     "Cell, Antibody +": {
-     "tp": 54,
-     "fp": 19,
-     "fn": 7,
-     "precision": 0.7397260273972602,
-     "recall": 0.8852459016393442,
-     "f1": 0.8059701492537313,
      "gt_count": 61,
-     "pred_count": 73
     },
     "Cell, Antibody -": {
      "tp": 1,
-     "fp": 0,
      "fn": 0,
-     "precision": 1.0,
      "recall": 1.0,
-     "f1": 1.0,
      "gt_count": 1,
-     "pred_count": 1
     },
     "Cell, Nostril, Antibody +": {
-     "tp": 102,
-     "fp": 10,
-     "fn": 20,
-     "precision": 0.9107142857142857,
-     "recall": 0.8360655737704918,
-     "f1": 0.8717948717948717,
      "gt_count": 122,
-     "pred_count": 112
     },
     "Cell, Nostril, Antibody -": {
      "tp": 0,
@@ -480,9 +480,9 @@
    },
    "confusion": {
     "Cell, Antibody +": {
-     "Cell, Antibody +": 54,
-     "Cell, Antibody -": 0,
-     "Cell, Nostril, Antibody +": 3,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Antibody -": {
@@ -492,9 +492,9 @@
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody +": {
-     "Cell, Antibody +": 4,
      "Cell, Antibody -": 0,
-     "Cell, Nostril, Antibody +": 102,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody -": {
@@ -505,62 +505,62 @@
     }
    },
    "n_gt": 184,
-   "n_pred": 186
   },
   "nostrils": {
    "detection": {
     "tp": 347,
-    "fp": 93,
     "fn": 29,
-    "precision": 0.7886363636363637,
     "recall": 0.9228723404255319,
-    "f1": 0.8504901960784313,
-    "mean_dist_matched": 0.8862671910950328
    },
-   "type_accuracy": 0.9798270893371758,
    "per_class": {
     "Nostril in Nucleus": {
-     "tp": 148,
-     "fp": 34,
-     "fn": 21,
-     "precision": 0.8131868131868132,
-     "recall": 0.8757396449704142,
-     "f1": 0.8433048433048432,
      "gt_count": 169,
-     "pred_count": 182
     },
     "Nostril External of Nucleus": {
-     "tp": 192,
-     "fp": 66,
-     "fn": 15,
-     "precision": 0.7441860465116279,
-     "recall": 0.927536231884058,
-     "f1": 0.8258064516129032,
      "gt_count": 207,
-     "pred_count": 258
     }
    },
    "n_gt": 376,
-   "n_pred": 440
   },
   "counts": {
    "Cell, Antibody +": {
     "gt": 61,
-    "pred": 73,
-    "abs_err": 12,
-    "rel_err": 0.19672131147540983
    },
    "Cell, Antibody -": {
     "gt": 1,
-    "pred": 1,
-    "abs_err": 0,
-    "rel_err": 0.0
    },
    "Cell, Nostril, Antibody +": {
     "gt": 122,
-    "pred": 112,
-    "abs_err": 10,
-    "rel_err": 0.08196721311475409
    },
    "Cell, Nostril, Antibody -": {
     "gt": 0,
@@ -570,104 +570,104 @@
    },
    "Nostril in Nucleus": {
     "gt": 169,
-    "pred": 182,
-    "abs_err": 13,
-    "rel_err": 0.07692307692307693
    },
    "Nostril External of Nucleus": {
     "gt": 207,
-    "pred": 258,
-    "abs_err": 51,
-    "rel_err": 0.2463768115942029
    }
   },
   "fp_analysis": null,
   "pairs": null,
   "summary": {
-   "cell_det_f1": 0.8864864864864864,
-   "cell_det_precision": 0.8817204301075269,
-   "cell_det_recall": 0.8913043478260869,
-   "cell_mean_iou": 0.800839882168248,
-   "cell_det_f1_iou30": 0.9351351351351351,
-   "cell_det_f1_centroid": 0.9297297297297297,
-   "cell_attr_antibody": 1.0,
-   "cell_attr_nostril": 0.9573170731707317,
-   "cell_attr_both": 0.9573170731707317,
-   "nos_det_f1": 0.8504901960784313,
-   "nos_det_precision": 0.7886363636363637,
    "nos_det_recall": 0.9228723404255319,
-   "nos_type_acc": 0.9798270893371758,
-   "f1[Cell, Antibody +]": 0.8059701492537313,
-   "f1[Cell, Antibody -]": 1.0,
-   "f1[Cell, Nostril, Antibody +]": 0.8717948717948717,
    "f1[Cell, Nostril, Antibody -]": 0.0,
-   "f1[Nostril in Nucleus]": 0.8433048433048432,
-   "f1[Nostril External of Nucleus]": 0.8258064516129032
   }
  },
  "C112": {
   "cells": {
    "detection": {
-    "tp": 124,
-    "fp": 26,
-    "fn": 22,
-    "precision": 0.8266666666666667,
-    "recall": 0.8493150684931506,
-    "f1": 0.8378378378378378,
-    "mean_iou_matched": 0.7526341825198912,
     "iou30": {
-     "tp": 134,
-     "fp": 16,
-     "fn": 12,
-     "precision": 0.8933333333333333,
-     "recall": 0.9178082191780822,
-     "f1": 0.9054054054054055
     },
     "centroid": {
-     "tp": 133,
-     "fp": 17,
-     "fn": 13,
-     "precision": 0.8866666666666667,
-     "recall": 0.910958904109589,
-     "f1": 0.8986486486486487
     }
    },
    "attribute_accuracy": {
-    "antibody": 0.9919354838709677,
-    "nostril": 1.0,
-    "both": 0.9919354838709677,
-    "n_matched": 124
    },
    "per_class": {
     "Cell, Antibody +": {
-     "tp": 104,
-     "fp": 22,
-     "fn": 20,
-     "precision": 0.8253968253968254,
-     "recall": 0.8387096774193549,
-     "f1": 0.832,
      "gt_count": 124,
-     "pred_count": 126
     },
     "Cell, Antibody -": {
-     "tp": 4,
      "fp": 1,
-     "fn": 1,
-     "precision": 0.8,
-     "recall": 0.8,
-     "f1": 0.8000000000000002,
      "gt_count": 5,
-     "pred_count": 5
     },
     "Cell, Nostril, Antibody +": {
      "tp": 15,
-     "fp": 4,
      "fn": 2,
-     "precision": 0.7894736842105263,
      "recall": 0.8823529411764706,
-     "f1": 0.8333333333333333,
      "gt_count": 17,
-     "pred_count": 19
     },
     "Cell, Nostril, Antibody -": {
      "tp": 0,
@@ -682,14 +682,14 @@
    },
    "confusion": {
     "Cell, Antibody +": {
-     "Cell, Antibody +": 104,
      "Cell, Antibody -": 0,
-     "Cell, Nostril, Antibody +": 0,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Antibody -": {
-     "Cell, Antibody +": 1,
-     "Cell, Antibody -": 4,
      "Cell, Nostril, Antibody +": 0,
      "Cell, Nostril, Antibody -": 0
     },
@@ -707,62 +707,62 @@
     }
    },
    "n_gt": 146,
-   "n_pred": 150
   },
   "nostrils": {
    "detection": {
     "tp": 25,
-    "fp": 8,
     "fn": 0,
-    "precision": 0.7575757575757576,
     "recall": 1.0,
-    "f1": 0.8620689655172413,
-    "mean_dist_matched": 0.934525036066851
    },
    "type_accuracy": 1.0,
    "per_class": {
     "Nostril in Nucleus": {
      "tp": 23,
-     "fp": 2,
      "fn": 0,
-     "precision": 0.92,
      "recall": 1.0,
-     "f1": 0.9583333333333334,
      "gt_count": 23,
-     "pred_count": 25
     },
     "Nostril External of Nucleus": {
      "tp": 2,
-     "fp": 6,
      "fn": 0,
-     "precision": 0.25,
      "recall": 1.0,
-     "f1": 0.4,
      "gt_count": 2,
-     "pred_count": 8
     }
    },
    "n_gt": 25,
-   "n_pred": 33
   },
   "counts": {
    "Cell, Antibody +": {
     "gt": 124,
-    "pred": 126,
-    "abs_err": 2,
-    "rel_err": 0.016129032258064516
    },
    "Cell, Antibody -": {
     "gt": 5,
-    "pred": 5,
-    "abs_err": 0,
-    "rel_err": 0.0
    },
    "Cell, Nostril, Antibody +": {
     "gt": 17,
-    "pred": 19,
-    "abs_err": 2,
-    "rel_err": 0.11764705882352941
    },
    "Cell, Nostril, Antibody -": {
     "gt": 0,
@@ -772,104 +772,104 @@
    },
    "Nostril in Nucleus": {
     "gt": 23,
-    "pred": 25,
-    "abs_err": 2,
-    "rel_err": 0.08695652173913043
    },
    "Nostril External of Nucleus": {
     "gt": 2,
-    "pred": 8,
-    "abs_err": 6,
-    "rel_err": 3.0
    }
   },
   "fp_analysis": null,
   "pairs": null,
   "summary": {
-   "cell_det_f1": 0.8378378378378378,
-   "cell_det_precision": 0.8266666666666667,
-   "cell_det_recall": 0.8493150684931506,
-   "cell_mean_iou": 0.7526341825198912,
-   "cell_det_f1_iou30": 0.9054054054054055,
-   "cell_det_f1_centroid": 0.8986486486486487,
-   "cell_attr_antibody": 0.9919354838709677,
-   "cell_attr_nostril": 1.0,
-   "cell_attr_both": 0.9919354838709677,
-   "nos_det_f1": 0.8620689655172413,
-   "nos_det_precision": 0.7575757575757576,
    "nos_det_recall": 1.0,
    "nos_type_acc": 1.0,
-   "f1[Cell, Antibody +]": 0.832,
-   "f1[Cell, Antibody -]": 0.8000000000000002,
-   "f1[Cell, Nostril, Antibody +]": 0.8333333333333333,
    "f1[Cell, Nostril, Antibody -]": 0.0,
-   "f1[Nostril in Nucleus]": 0.9583333333333334,
-   "f1[Nostril External of Nucleus]": 0.4
   }
  },
  "F100": {
   "cells": {
    "detection": {
-    "tp": 160,
-    "fp": 21,
-    "fn": 26,
-    "precision": 0.8839779005524862,
-    "recall": 0.8602150537634409,
-    "f1": 0.871934604904632,
-    "mean_iou_matched": 0.8049454720324757,
     "iou30": {
-     "tp": 169,
-     "fp": 12,
-     "fn": 17,
-     "precision": 0.9337016574585635,
-     "recall": 0.9086021505376344,
-     "f1": 0.9209809264305177
     },
     "centroid": {
-     "tp": 168,
-     "fp": 13,
-     "fn": 18,
-     "precision": 0.9281767955801105,
-     "recall": 0.9032258064516129,
-     "f1": 0.9155313351498637
     }
    },
    "attribute_accuracy": {
-    "antibody": 0.98125,
-    "nostril": 0.98125,
-    "both": 0.9625,
-    "n_matched": 160
    },
    "per_class": {
     "Cell, Antibody +": {
-     "tp": 79,
-     "fp": 17,
-     "fn": 8,
-     "precision": 0.8229166666666666,
-     "recall": 0.9080459770114943,
-     "f1": 0.8633879781420765,
      "gt_count": 87,
-     "pred_count": 96
     },
     "Cell, Antibody -": {
-     "tp": 4,
-     "fp": 2,
-     "fn": 2,
-     "precision": 0.6666666666666666,
-     "recall": 0.6666666666666666,
-     "f1": 0.6666666666666666,
      "gt_count": 6,
-     "pred_count": 6
     },
     "Cell, Nostril, Antibody +": {
-     "tp": 71,
-     "fp": 8,
-     "fn": 22,
-     "precision": 0.8987341772151899,
-     "recall": 0.7634408602150538,
-     "f1": 0.8255813953488371,
      "gt_count": 93,
-     "pred_count": 79
     },
     "Cell, Nostril, Antibody -": {
      "tp": 0,
@@ -884,21 +884,21 @@
    },
    "confusion": {
     "Cell, Antibody +": {
-     "Cell, Antibody +": 79,
      "Cell, Antibody -": 1,
-     "Cell, Nostril, Antibody +": 0,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Antibody -": {
-     "Cell, Antibody +": 2,
-     "Cell, Antibody -": 4,
      "Cell, Nostril, Antibody +": 0,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody +": {
-     "Cell, Antibody +": 3,
      "Cell, Antibody -": 0,
-     "Cell, Nostril, Antibody +": 71,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody -": {
@@ -909,62 +909,62 @@
     }
    },
    "n_gt": 186,
-   "n_pred": 181
   },
   "nostrils": {
    "detection": {
-    "tp": 142,
-    "fp": 9,
-    "fn": 20,
-    "precision": 0.9403973509933775,
-    "recall": 0.8765432098765432,
-    "f1": 0.9073482428115016,
-    "mean_dist_matched": 1.2296582069626907
-   },
-   "type_accuracy": 0.9929577464788732,
    "per_class": {
     "Nostril in Nucleus": {
-     "tp": 102,
-     "fp": 8,
-     "fn": 16,
-     "precision": 0.9272727272727272,
-     "recall": 0.864406779661017,
-     "f1": 0.8947368421052632,
      "gt_count": 118,
-     "pred_count": 110
     },
     "Nostril External of Nucleus": {
-     "tp": 39,
-     "fp": 2,
-     "fn": 5,
-     "precision": 0.9512195121951219,
-     "recall": 0.8863636363636364,
-     "f1": 0.9176470588235294,
      "gt_count": 44,
      "pred_count": 41
     }
    },
    "n_gt": 162,
-   "n_pred": 151
   },
   "counts": {
    "Cell, Antibody +": {
     "gt": 87,
-    "pred": 96,
-    "abs_err": 9,
-    "rel_err": 0.10344827586206896
    },
    "Cell, Antibody -": {
     "gt": 6,
-    "pred": 6,
-    "abs_err": 0,
-    "rel_err": 0.0
    },
    "Cell, Nostril, Antibody +": {
     "gt": 93,
-    "pred": 79,
-    "abs_err": 14,
-    "rel_err": 0.15053763440860216
    },
    "Cell, Nostril, Antibody -": {
     "gt": 0,
@@ -974,9 +974,9 @@
    },
    "Nostril in Nucleus": {
     "gt": 118,
-    "pred": 110,
-    "abs_err": 8,
-    "rel_err": 0.06779661016949153
    },
    "Nostril External of Nucleus": {
     "gt": 44,
@@ -988,90 +988,90 @@
   "fp_analysis": null,
   "pairs": null,
   "summary": {
-   "cell_det_f1": 0.871934604904632,
-   "cell_det_precision": 0.8839779005524862,
-   "cell_det_recall": 0.8602150537634409,
-   "cell_mean_iou": 0.8049454720324757,
-   "cell_det_f1_iou30": 0.9209809264305177,
-   "cell_det_f1_centroid": 0.9155313351498637,
-   "cell_attr_antibody": 0.98125,
-   "cell_attr_nostril": 0.98125,
-   "cell_attr_both": 0.9625,
-   "nos_det_f1": 0.9073482428115016,
-   "nos_det_precision": 0.9403973509933775,
-   "nos_det_recall": 0.8765432098765432,
-   "nos_type_acc": 0.9929577464788732,
-   "f1[Cell, Antibody +]": 0.8633879781420765,
-   "f1[Cell, Antibody -]": 0.6666666666666666,
-   "f1[Cell, Nostril, Antibody +]": 0.8255813953488371,
    "f1[Cell, Nostril, Antibody -]": 0.0,
-   "f1[Nostril in Nucleus]": 0.8947368421052632,
-   "f1[Nostril External of Nucleus]": 0.9176470588235294
   }
  },
  "F110": {
   "cells": {
    "detection": {
-    "tp": 116,
-    "fp": 9,
-    "fn": 12,
-    "precision": 0.928,
-    "recall": 0.90625,
-    "f1": 0.9169960474308302,
-    "mean_iou_matched": 0.8161455695675424,
     "iou30": {
-     "tp": 120,
-     "fp": 5,
-     "fn": 8,
-     "precision": 0.96,
-     "recall": 0.9375,
-     "f1": 0.9486166007905138
     },
     "centroid": {
-     "tp": 121,
-     "fp": 4,
-     "fn": 7,
-     "precision": 0.968,
-     "recall": 0.9453125,
-     "f1": 0.9565217391304348
     }
    },
    "attribute_accuracy": {
-    "antibody": 0.9827586206896551,
-    "nostril": 0.9655172413793104,
-    "both": 0.9482758620689655,
-    "n_matched": 116
    },
    "per_class": {
     "Cell, Antibody +": {
-     "tp": 82,
-     "fp": 10,
-     "fn": 5,
-     "precision": 0.8913043478260869,
-     "recall": 0.9425287356321839,
-     "f1": 0.9162011173184358,
      "gt_count": 87,
-     "pred_count": 92
     },
     "Cell, Antibody -": {
-     "tp": 5,
-     "fp": 0,
-     "fn": 2,
-     "precision": 1.0,
-     "recall": 0.7142857142857143,
-     "f1": 0.8333333333333333,
      "gt_count": 7,
-     "pred_count": 5
     },
     "Cell, Nostril, Antibody +": {
-     "tp": 23,
      "fp": 5,
-     "fn": 10,
-     "precision": 0.8214285714285714,
-     "recall": 0.696969696969697,
-     "f1": 0.7540983606557378,
      "gt_count": 33,
-     "pred_count": 28
     },
     "Cell, Nostril, Antibody -": {
      "tp": 0,
@@ -1086,21 +1086,21 @@
    },
    "confusion": {
     "Cell, Antibody +": {
-     "Cell, Antibody +": 82,
      "Cell, Antibody -": 0,
      "Cell, Nostril, Antibody +": 0,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Antibody -": {
-     "Cell, Antibody +": 2,
-     "Cell, Antibody -": 5,
      "Cell, Nostril, Antibody +": 0,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody +": {
-     "Cell, Antibody +": 4,
      "Cell, Antibody -": 0,
-     "Cell, Nostril, Antibody +": 23,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody -": {
@@ -1111,62 +1111,62 @@
     }
    },
    "n_gt": 128,
-   "n_pred": 125
   },
   "nostrils": {
    "detection": {
-    "tp": 157,
-    "fp": 9,
-    "fn": 37,
-    "precision": 0.9457831325301205,
-    "recall": 0.8092783505154639,
-    "f1": 0.8722222222222222,
-    "mean_dist_matched": 0.7899268230101212
-   },
-   "type_accuracy": 0.9936305732484076,
    "per_class": {
     "Nostril in Nucleus": {
-     "tp": 48,
-     "fp": 4,
-     "fn": 9,
-     "precision": 0.9230769230769231,
-     "recall": 0.8421052631578947,
-     "f1": 0.8807339449541285,
      "gt_count": 57,
-     "pred_count": 52
     },
     "Nostril External of Nucleus": {
-     "tp": 108,
-     "fp": 6,
-     "fn": 29,
-     "precision": 0.9473684210526315,
-     "recall": 0.7883211678832117,
-     "f1": 0.8605577689243028,
      "gt_count": 137,
-     "pred_count": 114
     }
    },
    "n_gt": 194,
-   "n_pred": 166
   },
   "counts": {
    "Cell, Antibody +": {
     "gt": 87,
-    "pred": 92,
-    "abs_err": 5,
-    "rel_err": 0.05747126436781609
    },
    "Cell, Antibody -": {
     "gt": 7,
-    "pred": 5,
-    "abs_err": 2,
-    "rel_err": 0.2857142857142857
    },
    "Cell, Nostril, Antibody +": {
     "gt": 33,
-    "pred": 28,
-    "abs_err": 5,
-    "rel_err": 0.15151515151515152
    },
    "Cell, Nostril, Antibody -": {
     "gt": 1,
@@ -1176,104 +1176,104 @@
    },
    "Nostril in Nucleus": {
     "gt": 57,
-    "pred": 52,
-    "abs_err": 5,
-    "rel_err": 0.08771929824561403
    },
    "Nostril External of Nucleus": {
     "gt": 137,
-    "pred": 114,
-    "abs_err": 23,
-    "rel_err": 0.1678832116788321
    }
   },
   "fp_analysis": null,
   "pairs": null,
   "summary": {
-   "cell_det_f1": 0.9169960474308302,
-   "cell_det_precision": 0.928,
-   "cell_det_recall": 0.90625,
-   "cell_mean_iou": 0.8161455695675424,
-   "cell_det_f1_iou30": 0.9486166007905138,
-   "cell_det_f1_centroid": 0.9565217391304348,
-   "cell_attr_antibody": 0.9827586206896551,
-   "cell_attr_nostril": 0.9655172413793104,
-   "cell_attr_both": 0.9482758620689655,
-   "nos_det_f1": 0.8722222222222222,
-   "nos_det_precision": 0.9457831325301205,
-   "nos_det_recall": 0.8092783505154639,
-   "nos_type_acc": 0.9936305732484076,
-   "f1[Cell, Antibody +]": 0.9162011173184358,
-   "f1[Cell, Antibody -]": 0.8333333333333333,
-   "f1[Cell, Nostril, Antibody +]": 0.7540983606557378,
    "f1[Cell, Nostril, Antibody -]": 0.0,
-   "f1[Nostril in Nucleus]": 0.8807339449541285,
-   "f1[Nostril External of Nucleus]": 0.8605577689243028
   }
  },
  "F113": {
   "cells": {
    "detection": {
-    "tp": 131,
-    "fp": 19,
-    "fn": 32,
-    "precision": 0.8733333333333333,
-    "recall": 0.803680981595092,
-    "f1": 0.8370607028753994,
-    "mean_iou_matched": 0.7799368783421308,
     "iou30": {
-     "tp": 146,
-     "fp": 4,
-     "fn": 17,
-     "precision": 0.9733333333333334,
-     "recall": 0.8957055214723927,
-     "f1": 0.9329073482428116
     },
     "centroid": {
-     "tp": 144,
-     "fp": 6,
-     "fn": 19,
-     "precision": 0.96,
-     "recall": 0.8834355828220859,
-     "f1": 0.9201277955271565
     }
    },
    "attribute_accuracy": {
     "antibody": 1.0,
-    "nostril": 0.9770992366412213,
-    "both": 0.9770992366412213,
-    "n_matched": 131
    },
    "per_class": {
     "Cell, Antibody +": {
-     "tp": 108,
-     "fp": 21,
-     "fn": 30,
-     "precision": 0.8372093023255814,
-     "recall": 0.782608695652174,
-     "f1": 0.8089887640449438,
      "gt_count": 138,
-     "pred_count": 129
     },
     "Cell, Antibody -": {
      "tp": 2,
-     "fp": 0,
      "fn": 1,
-     "precision": 1.0,
      "recall": 0.6666666666666666,
-     "f1": 0.8,
      "gt_count": 3,
-     "pred_count": 2
     },
     "Cell, Nostril, Antibody +": {
      "tp": 18,
-     "fp": 1,
      "fn": 4,
-     "precision": 0.9473684210526315,
      "recall": 0.8181818181818182,
-     "f1": 0.8780487804878049,
      "gt_count": 22,
-     "pred_count": 19
     },
     "Cell, Nostril, Antibody -": {
      "tp": 0,
@@ -1288,7 +1288,7 @@
    },
    "confusion": {
     "Cell, Antibody +": {
-     "Cell, Antibody +": 108,
      "Cell, Antibody -": 0,
      "Cell, Nostril, Antibody +": 0,
      "Cell, Nostril, Antibody -": 0
@@ -1313,62 +1313,62 @@
     }
    },
    "n_gt": 163,
-   "n_pred": 150
   },
   "nostrils": {
    "detection": {
-    "tp": 27,
-    "fp": 1,
-    "fn": 7,
-    "precision": 0.9642857142857143,
-    "recall": 0.7941176470588235,
-    "f1": 0.8709677419354839,
-    "mean_dist_matched": 0.8145690279889126
-   },
-   "type_accuracy": 0.9259259259259259,
    "per_class": {
     "Nostril in Nucleus": {
-     "tp": 21,
-     "fp": 2,
-     "fn": 5,
-     "precision": 0.9130434782608695,
-     "recall": 0.8076923076923077,
-     "f1": 0.8571428571428572,
      "gt_count": 26,
      "pred_count": 23
     },
     "Nostril External of Nucleus": {
-     "tp": 4,
      "fp": 1,
-     "fn": 4,
-     "precision": 0.8,
-     "recall": 0.5,
-     "f1": 0.6153846153846154,
      "gt_count": 8,
-     "pred_count": 5
     }
    },
    "n_gt": 34,
-   "n_pred": 28
   },
   "counts": {
    "Cell, Antibody +": {
     "gt": 138,
-    "pred": 129,
-    "abs_err": 9,
-    "rel_err": 0.06521739130434782
    },
    "Cell, Antibody -": {
     "gt": 3,
-    "pred": 2,
-    "abs_err": 1,
-    "rel_err": 0.3333333333333333
    },
    "Cell, Nostril, Antibody +": {
     "gt": 22,
-    "pred": 19,
-    "abs_err": 3,
-    "rel_err": 0.13636363636363635
    },
    "Cell, Nostril, Antibody -": {
     "gt": 0,
@@ -1384,298 +1384,298 @@
    },
    "Nostril External of Nucleus": {
     "gt": 8,
-    "pred": 5,
-    "abs_err": 3,
-    "rel_err": 0.375
    }
   },
   "fp_analysis": null,
   "pairs": null,
   "summary": {
-   "cell_det_f1": 0.8370607028753994,
-   "cell_det_precision": 0.8733333333333333,
-   "cell_det_recall": 0.803680981595092,
-   "cell_mean_iou": 0.7799368783421308,
-   "cell_det_f1_iou30": 0.9329073482428116,
-   "cell_det_f1_centroid": 0.9201277955271565,
    "cell_attr_antibody": 1.0,
-   "cell_attr_nostril": 0.9770992366412213,
-   "cell_attr_both": 0.9770992366412213,
-   "nos_det_f1": 0.8709677419354839,
-   "nos_det_precision": 0.9642857142857143,
-   "nos_det_recall": 0.7941176470588235,
-   "nos_type_acc": 0.9259259259259259,
-   "f1[Cell, Antibody +]": 0.8089887640449438,
-   "f1[Cell, Antibody -]": 0.8,
-   "f1[Cell, Nostril, Antibody +]": 0.8780487804878049,
    "f1[Cell, Nostril, Antibody -]": 0.0,
-   "f1[Nostril in Nucleus]": 0.8571428571428572,
-   "f1[Nostril External of Nucleus]": 0.6153846153846154
   }
  },
  "F114": {
   "cells": {
    "detection": {
-    "tp": 211,
-    "fp": 25,
-    "fn": 31,
-    "precision": 0.8940677966101694,
-    "recall": 0.871900826446281,
-    "f1": 0.8828451882845189,
-    "mean_iou_matched": 0.8168843572713155,
     "iou30": {
-     "tp": 222,
-     "fp": 14,
-     "fn": 20,
-     "precision": 0.940677966101695,
-     "recall": 0.9173553719008265,
-     "f1": 0.9288702928870294
     },
     "centroid": {
-     "tp": 221,
-     "fp": 15,
-     "fn": 21,
-     "precision": 0.9364406779661016,
-     "recall": 0.9132231404958677,
-     "f1": 0.9246861924686192
     }
    },
    "attribute_accuracy": {
-    "antibody": 0.95260663507109,
-    "nostril": 0.981042654028436,
-    "both": 0.933649289099526,
-    "n_matched": 211
    },
    "per_class": {
     "Cell, Antibody +": {
-     "tp": 148,
-     "fp": 28,
-     "fn": 27,
-     "precision": 0.8409090909090909,
-     "recall": 0.8457142857142858,
-     "f1": 0.8433048433048433,
      "gt_count": 175,
-     "pred_count": 176
     },
     "Cell, Antibody -": {
-     "tp": 12,
-     "fp": 2,
-     "fn": 9,
-     "precision": 0.8571428571428571,
-     "recall": 0.5714285714285714,
-     "f1": 0.6857142857142857,
      "gt_count": 21,
-     "pred_count": 14
     },
     "Cell, Nostril, Antibody +": {
-     "tp": 37,
-     "fp": 9,
-     "fn": 5,
-     "precision": 0.8043478260869565,
-     "recall": 0.8809523809523809,
-     "f1": 0.8409090909090908,
      "gt_count": 42,
-     "pred_count": 46
     },
     "Cell, Nostril, Antibody -": {
-     "tp": 0,
      "fp": 0,
-     "fn": 4,
-     "precision": 0.0,
-     "recall": 0.0,
-     "f1": 0.0,
      "gt_count": 4,
-     "pred_count": 0
     }
    },
    "confusion": {
     "Cell, Antibody +": {
-     "Cell, Antibody +": 148,
-     "Cell, Antibody -": 1,
-     "Cell, Nostril, Antibody +": 4,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Antibody -": {
-     "Cell, Antibody +": 9,
-     "Cell, Antibody -": 12,
      "Cell, Nostril, Antibody +": 0,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody +": {
-     "Cell, Antibody +": 0,
      "Cell, Antibody -": 0,
-     "Cell, Nostril, Antibody +": 37,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody -": {
      "Cell, Antibody +": 0,
-     "Cell, Antibody -": 0,
      "Cell, Nostril, Antibody +": 0,
-     "Cell, Nostril, Antibody -": 0
     }
    },
    "n_gt": 242,
-   "n_pred": 236
   },
   "nostrils": {
    "detection": {
-    "tp": 81,
-    "fp": 24,
-    "fn": 9,
-    "precision": 0.7714285714285715,
-    "recall": 0.9,
-    "f1": 0.8307692307692307,
-    "mean_dist_matched": 0.9417483459046142
    },
    "type_accuracy": 1.0,
    "per_class": {
     "Nostril in Nucleus": {
-     "tp": 46,
-     "fp": 8,
-     "fn": 6,
-     "precision": 0.8518518518518519,
-     "recall": 0.8846153846153846,
-     "f1": 0.8679245283018868,
      "gt_count": 52,
-     "pred_count": 54
     },
     "Nostril External of Nucleus": {
-     "tp": 35,
-     "fp": 16,
-     "fn": 3,
-     "precision": 0.6862745098039216,
-     "recall": 0.9210526315789473,
-     "f1": 0.7865168539325842,
      "gt_count": 38,
-     "pred_count": 51
     }
    },
    "n_gt": 90,
-   "n_pred": 105
   },
   "counts": {
    "Cell, Antibody +": {
     "gt": 175,
-    "pred": 176,
-    "abs_err": 1,
-    "rel_err": 0.005714285714285714
    },
    "Cell, Antibody -": {
     "gt": 21,
-    "pred": 14,
-    "abs_err": 7,
-    "rel_err": 0.3333333333333333
    },
    "Cell, Nostril, Antibody +": {
     "gt": 42,
-    "pred": 46,
-    "abs_err": 4,
-    "rel_err": 0.09523809523809523
    },
    "Cell, Nostril, Antibody -": {
     "gt": 4,
-    "pred": 0,
-    "abs_err": 4,
-    "rel_err": 1.0
    },
    "Nostril in Nucleus": {
     "gt": 52,
-    "pred": 54,
-    "abs_err": 2,
-    "rel_err": 0.038461538461538464
    },
    "Nostril External of Nucleus": {
     "gt": 38,
-    "pred": 51,
-    "abs_err": 13,
-    "rel_err": 0.34210526315789475
    }
   },
   "fp_analysis": null,
   "pairs": null,
   "summary": {
-   "cell_det_f1": 0.8828451882845189,
-   "cell_det_precision": 0.8940677966101694,
-   "cell_det_recall": 0.871900826446281,
-   "cell_mean_iou": 0.8168843572713155,
-   "cell_det_f1_iou30": 0.9288702928870294,
-   "cell_det_f1_centroid": 0.9246861924686192,
-   "cell_attr_antibody": 0.95260663507109,
-   "cell_attr_nostril": 0.981042654028436,
-   "cell_attr_both": 0.933649289099526,
-   "nos_det_f1": 0.8307692307692307,
-   "nos_det_precision": 0.7714285714285715,
-   "nos_det_recall": 0.9,
    "nos_type_acc": 1.0,
-   "f1[Cell, Antibody +]": 0.8433048433048433,
-   "f1[Cell, Antibody -]": 0.6857142857142857,
-   "f1[Cell, Nostril, Antibody +]": 0.8409090909090908,
-   "f1[Cell, Nostril, Antibody -]": 0.0,
-   "f1[Nostril in Nucleus]": 0.8679245283018868,
-   "f1[Nostril External of Nucleus]": 0.7865168539325842
   }
  },
  "F99": {
   "cells": {
    "detection": {
-    "tp": 227,
-    "fp": 28,
-    "fn": 24,
-    "precision": 0.8901960784313725,
-    "recall": 0.9043824701195219,
-    "f1": 0.8972332015810275,
-    "mean_iou_matched": 0.8280746396024866,
     "iou30": {
-     "tp": 238,
-     "fp": 17,
-     "fn": 13,
-     "precision": 0.9333333333333333,
-     "recall": 0.9482071713147411,
-     "f1": 0.9407114624505929
     },
     "centroid": {
-     "tp": 238,
-     "fp": 17,
-     "fn": 13,
-     "precision": 0.9333333333333333,
-     "recall": 0.9482071713147411,
-     "f1": 0.9407114624505929
     }
    },
    "attribute_accuracy": {
-    "antibody": 0.920704845814978,
-    "nostril": 0.9911894273127754,
-    "both": 0.9118942731277533,
-    "n_matched": 227
    },
    "per_class": {
     "Cell, Antibody +": {
-     "tp": 162,
-     "fp": 41,
-     "fn": 21,
-     "precision": 0.7980295566502463,
-     "recall": 0.8852459016393442,
-     "f1": 0.839378238341969,
      "gt_count": 183,
-     "pred_count": 203
     },
     "Cell, Antibody -": {
-     "tp": 42,
-     "fp": 6,
-     "fn": 20,
-     "precision": 0.875,
-     "recall": 0.6774193548387096,
-     "f1": 0.7636363636363636,
      "gt_count": 62,
-     "pred_count": 48
     },
     "Cell, Nostril, Antibody +": {
-     "tp": 3,
-     "fp": 1,
-     "fn": 3,
-     "precision": 0.75,
-     "recall": 0.5,
-     "f1": 0.6,
      "gt_count": 6,
      "pred_count": 4
     },
@@ -1692,21 +1692,21 @@
    },
    "confusion": {
     "Cell, Antibody +": {
-     "Cell, Antibody +": 162,
      "Cell, Antibody -": 4,
      "Cell, Nostril, Antibody +": 0,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Antibody -": {
-     "Cell, Antibody +": 14,
-     "Cell, Antibody -": 42,
      "Cell, Nostril, Antibody +": 0,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody +": {
      "Cell, Antibody +": 2,
      "Cell, Antibody -": 0,
-     "Cell, Nostril, Antibody +": 3,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody -": {
@@ -1717,17 +1717,17 @@
     }
    },
    "n_gt": 251,
-   "n_pred": 255
   },
   "nostrils": {
    "detection": {
-    "tp": 14,
-    "fp": 1,
-    "fn": 8,
-    "precision": 0.9333333333333333,
-    "recall": 0.6363636363636364,
-    "f1": 0.7567567567567568,
-    "mean_dist_matched": 0.6631874114708707
    },
    "type_accuracy": 1.0,
    "per_class": {
@@ -1742,31 +1742,31 @@
      "pred_count": 4
     },
     "Nostril External of Nucleus": {
-     "tp": 10,
-     "fp": 1,
-     "fn": 6,
-     "precision": 0.9090909090909091,
-     "recall": 0.625,
-     "f1": 0.7407407407407406,
      "gt_count": 16,
-     "pred_count": 11
     }
    },
    "n_gt": 22,
-   "n_pred": 15
   },
   "counts": {
    "Cell, Antibody +": {
     "gt": 183,
-    "pred": 203,
-    "abs_err": 20,
-    "rel_err": 0.1092896174863388
    },
    "Cell, Antibody -": {
     "gt": 62,
-    "pred": 48,
-    "abs_err": 14,
-    "rel_err": 0.22580645161290322
    },
    "Cell, Nostril, Antibody +": {
     "gt": 6,
@@ -1788,33 +1788,33 @@
    },
    "Nostril External of Nucleus": {
     "gt": 16,
-    "pred": 11,
-    "abs_err": 5,
-    "rel_err": 0.3125
    }
   },
   "fp_analysis": null,
   "pairs": null,
   "summary": {
-   "cell_det_f1": 0.8972332015810275,
-   "cell_det_precision": 0.8901960784313725,
-   "cell_det_recall": 0.9043824701195219,
-   "cell_mean_iou": 0.8280746396024866,
-   "cell_det_f1_iou30": 0.9407114624505929,
-   "cell_det_f1_centroid": 0.9407114624505929,
-   "cell_attr_antibody": 0.920704845814978,
-   "cell_attr_nostril": 0.9911894273127754,
-   "cell_attr_both": 0.9118942731277533,
-   "nos_det_f1": 0.7567567567567568,
-   "nos_det_precision": 0.9333333333333333,
-   "nos_det_recall": 0.6363636363636364,
    "nos_type_acc": 1.0,
-   "f1[Cell, Antibody +]": 0.839378238341969,
-   "f1[Cell, Antibody -]": 0.7636363636363636,
-   "f1[Cell, Nostril, Antibody +]": 0.6,
    "f1[Cell, Nostril, Antibody -]": 0.0,
    "f1[Nostril in Nucleus]": 0.8,
-   "f1[Nostril External of Nucleus]": 0.7407407407407406
   }
  }
 }

  "C100": {
   "cells": {
    "detection": {
+    "tp": 173,
+    "fp": 21,
+    "fn": 5,
+    "precision": 0.8917525773195877,
+    "recall": 0.9719101123595506,
+    "f1": 0.9301075268817205,
+    "mean_iou_matched": 0.795722646954622,
     "iou30": {
+     "tp": 177,
+     "fp": 17,
+     "fn": 1,
+     "precision": 0.9123711340206185,
+     "recall": 0.9943820224719101,
+     "f1": 0.9516129032258065
     },
     "centroid": {
+     "tp": 178,
+     "fp": 16,
+     "fn": 0,
+     "precision": 0.9175257731958762,
+     "recall": 1.0,
+     "f1": 0.956989247311828
     }
    },
    "attribute_accuracy": {
+    "antibody": 0.9884393063583815,
+    "nostril": 0.9942196531791907,
+    "both": 0.9826589595375722,
+    "n_matched": 173
    },
    "per_class": {
     "Cell, Antibody +": {
+     "tp": 158,
+     "fp": 18,
+     "fn": 7,
+     "precision": 0.8977272727272727,
+     "recall": 0.9575757575757575,
+     "f1": 0.9266862170087976,
      "gt_count": 165,
+     "pred_count": 176
     },
     "Cell, Antibody -": {
      "tp": 0,
+     "fp": 6,
      "fn": 0,
      "precision": 0.0,
      "recall": 0.0,
      "f1": 0.0,
      "gt_count": 0,
+     "pred_count": 6
     },
     "Cell, Nostril, Antibody +": {
      "tp": 12,
    },
    "confusion": {
     "Cell, Antibody +": {
+     "Cell, Antibody +": 158,
+     "Cell, Antibody -": 2,
      "Cell, Nostril, Antibody +": 0,
      "Cell, Nostril, Antibody -": 0
     },
     }
    },
    "n_gt": 178,
+   "n_pred": 194
   },
   "nostrils": {
    "detection": {
     "tp": 16,
+    "fp": 4,
     "fn": 4,
+    "precision": 0.8,
     "recall": 0.8,
+    "f1": 0.8000000000000002,
+    "mean_dist_matched": 1.1705934762183428
    },
    "type_accuracy": 1.0,
    "per_class": {
     "Nostril in Nucleus": {
      "tp": 12,
+     "fp": 1,
      "fn": 2,
+     "precision": 0.9230769230769231,
      "recall": 0.8571428571428571,
+     "f1": 0.888888888888889,
      "gt_count": 14,
+     "pred_count": 13
     },
     "Nostril External of Nucleus": {
      "tp": 4,
     }
    },
    "n_gt": 20,
+   "n_pred": 20
   },
   "counts": {
    "Cell, Antibody +": {
     "gt": 165,
+    "pred": 176,
+    "abs_err": 11,
+    "rel_err": 0.06666666666666667
    },
    "Cell, Antibody -": {
     "gt": 0,
+    "pred": 6,
+    "abs_err": 6,
     "rel_err": Infinity
    },
    "Cell, Nostril, Antibody +": {
    },
    "Nostril in Nucleus": {
     "gt": 14,
+    "pred": 13,
+    "abs_err": 1,
+    "rel_err": 0.07142857142857142
    },
    "Nostril External of Nucleus": {
     "gt": 6,
   "fp_analysis": null,
   "pairs": null,
   "summary": {
+   "cell_det_f1": 0.9301075268817205,
+   "cell_det_precision": 0.8917525773195877,
+   "cell_det_recall": 0.9719101123595506,
+   "cell_mean_iou": 0.795722646954622,
+   "cell_det_f1_iou30": 0.9516129032258065,
+   "cell_det_f1_centroid": 0.956989247311828,
+   "cell_attr_antibody": 0.9884393063583815,
+   "cell_attr_nostril": 0.9942196531791907,
+   "cell_attr_both": 0.9826589595375722,
+   "nos_det_f1": 0.8000000000000002,
+   "nos_det_precision": 0.8,
    "nos_det_recall": 0.8,
    "nos_type_acc": 1.0,
+   "f1[Cell, Antibody +]": 0.9266862170087976,
    "f1[Cell, Antibody -]": 0.0,
    "f1[Cell, Nostril, Antibody +]": 0.9600000000000001,
    "f1[Cell, Nostril, Antibody -]": 0.0,
+   "f1[Nostril in Nucleus]": 0.888888888888889,
    "f1[Nostril External of Nucleus]": 0.6153846153846153
   }
  },
  "C102": {
   "cells": {
    "detection": {
+    "tp": 91,
+    "fp": 17,
+    "fn": 0,
+    "precision": 0.8425925925925926,
+    "recall": 1.0,
+    "f1": 0.914572864321608,
+    "mean_iou_matched": 0.8036863197012898,
     "iou30": {
+     "tp": 91,
+     "fp": 17,
+     "fn": 0,
+     "precision": 0.8425925925925926,
+     "recall": 1.0,
+     "f1": 0.914572864321608
     },
     "centroid": {
+     "tp": 91,
+     "fp": 17,
+     "fn": 0,
+     "precision": 0.8425925925925926,
+     "recall": 1.0,
+     "f1": 0.914572864321608
     }
    },
    "attribute_accuracy": {
     "antibody": 1.0,
+    "nostril": 0.945054945054945,
+    "both": 0.945054945054945,
+    "n_matched": 91
    },
    "per_class": {
     "Cell, Antibody +": {
      "tp": 41,
+     "fp": 17,
      "fn": 0,
+     "precision": 0.7068965517241379,
      "recall": 1.0,
+     "f1": 0.8282828282828283,
      "gt_count": 41,
+     "pred_count": 58
     },
     "Cell, Antibody -": {
      "tp": 0,
+     "fp": 1,
      "fn": 0,
      "precision": 0.0,
      "recall": 0.0,
      "f1": 0.0,
      "gt_count": 0,
+     "pred_count": 1
     },
     "Cell, Nostril, Antibody +": {
+     "tp": 45,
+     "fp": 4,
+     "fn": 5,
+     "precision": 0.9183673469387755,
+     "recall": 0.9,
+     "f1": 0.9090909090909091,
      "gt_count": 50,
+     "pred_count": 49
     },
     "Cell, Nostril, Antibody -": {
      "tp": 0,
     "Cell, Nostril, Antibody +": {
      "Cell, Antibody +": 5,
      "Cell, Antibody -": 0,
+     "Cell, Nostril, Antibody +": 45,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody -": {
     }
    },
    "n_gt": 91,
+   "n_pred": 108
   },
   "nostrils": {
    "detection": {
+    "tp": 123,
+    "fp": 55,
+    "fn": 11,
+    "precision": 0.6910112359550562,
+    "recall": 0.917910447761194,
+    "f1": 0.7884615384615384,
+    "mean_dist_matched": 1.022617546328301
+   },
+   "type_accuracy": 0.991869918699187,
    "per_class": {
     "Nostril in Nucleus": {
      "tp": 54,
+     "fp": 13,
      "fn": 6,
+     "precision": 0.8059701492537313,
      "recall": 0.9,
+     "f1": 0.8503937007874016,
      "gt_count": 60,
+     "pred_count": 67
     },
     "Nostril External of Nucleus": {
+     "tp": 68,
+     "fp": 43,
+     "fn": 6,
+     "precision": 0.6126126126126126,
+     "recall": 0.918918918918919,
+     "f1": 0.7351351351351352,
      "gt_count": 74,
+     "pred_count": 111
     }
    },
    "n_gt": 134,
+   "n_pred": 178
   },
   "counts": {
    "Cell, Antibody +": {
     "gt": 41,
+    "pred": 58,
+    "abs_err": 17,
+    "rel_err": 0.4146341463414634
    },
    "Cell, Antibody -": {
     "gt": 0,
+    "pred": 1,
+    "abs_err": 1,
+    "rel_err": Infinity
    },
    "Cell, Nostril, Antibody +": {
     "gt": 50,
+    "pred": 49,
+    "abs_err": 1,
+    "rel_err": 0.02
    },
    "Cell, Nostril, Antibody -": {
     "gt": 0,
    },
    "Nostril in Nucleus": {
     "gt": 60,
+    "pred": 67,
+    "abs_err": 7,
+    "rel_err": 0.11666666666666667
    },
    "Nostril External of Nucleus": {
     "gt": 74,
+    "pred": 111,
+    "abs_err": 37,
+    "rel_err": 0.5
    }
   },
   "fp_analysis": null,
   "pairs": null,
   "summary": {
+   "cell_det_f1": 0.914572864321608,
+   "cell_det_precision": 0.8425925925925926,
+   "cell_det_recall": 1.0,
+   "cell_mean_iou": 0.8036863197012898,
+   "cell_det_f1_iou30": 0.914572864321608,
+   "cell_det_f1_centroid": 0.914572864321608,
    "cell_attr_antibody": 1.0,
+   "cell_attr_nostril": 0.945054945054945,
+   "cell_attr_both": 0.945054945054945,
+   "nos_det_f1": 0.7884615384615384,
+   "nos_det_precision": 0.6910112359550562,
+   "nos_det_recall": 0.917910447761194,
+   "nos_type_acc": 0.991869918699187,
+   "f1[Cell, Antibody +]": 0.8282828282828283,
    "f1[Cell, Antibody -]": 0.0,
+   "f1[Cell, Nostril, Antibody +]": 0.9090909090909091,
    "f1[Cell, Nostril, Antibody -]": 0.0,
+   "f1[Nostril in Nucleus]": 0.8503937007874016,
+   "f1[Nostril External of Nucleus]": 0.7351351351351352
   }
  },
  "C110": {
   "cells": {
    "detection": {
+    "tp": 168,
+    "fp": 34,
+    "fn": 16,
+    "precision": 0.8316831683168316,
+    "recall": 0.9130434782608695,
+    "f1": 0.8704663212435233,
+    "mean_iou_matched": 0.7832369871445182,
     "iou30": {
+     "tp": 178,
+     "fp": 24,
+     "fn": 6,
+     "precision": 0.8811881188118812,
+     "recall": 0.967391304347826,
+     "f1": 0.922279792746114
     },
     "centroid": {
+     "tp": 180,
+     "fp": 22,
+     "fn": 4,
+     "precision": 0.8910891089108911,
+     "recall": 0.9782608695652174,
+     "f1": 0.9326424870466321
     }
    },
    "attribute_accuracy": {
+    "antibody": 0.9940476190476191,
+    "nostril": 0.9702380952380952,
+    "both": 0.9642857142857143,
+    "n_matched": 168
    },
    "per_class": {
     "Cell, Antibody +": {
+     "tp": 52,
+     "fp": 23,
+     "fn": 9,
+     "precision": 0.6933333333333334,
+     "recall": 0.8524590163934426,
+     "f1": 0.7647058823529411,
      "gt_count": 61,
+     "pred_count": 75
     },
     "Cell, Antibody -": {
      "tp": 1,
+     "fp": 1,
      "fn": 0,
+     "precision": 0.5,
      "recall": 1.0,
+     "f1": 0.6666666666666666,
      "gt_count": 1,
+     "pred_count": 2
     },
     "Cell, Nostril, Antibody +": {
+     "tp": 109,
+     "fp": 16,
+     "fn": 13,
+     "precision": 0.872,
+     "recall": 0.8934426229508197,
+     "f1": 0.8825910931174089,
      "gt_count": 122,
+     "pred_count": 125
     },
     "Cell, Nostril, Antibody -": {
      "tp": 0,
    },
    "confusion": {
     "Cell, Antibody +": {
+     "Cell, Antibody +": 52,
+     "Cell, Antibody -": 1,
+     "Cell, Nostril, Antibody +": 2,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Antibody -": {
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody +": {
+     "Cell, Antibody +": 3,
      "Cell, Antibody -": 0,
+     "Cell, Nostril, Antibody +": 109,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody -": {
     }
    },
    "n_gt": 184,
+   "n_pred": 202
   },
   "nostrils": {
    "detection": {
     "tp": 347,
+    "fp": 104,
     "fn": 29,
+    "precision": 0.7694013303769401,
     "recall": 0.9228723404255319,
+    "f1": 0.8391777509068925,
+    "mean_dist_matched": 0.9012807114492273
    },
+   "type_accuracy": 0.9740634005763689,
    "per_class": {
     "Nostril in Nucleus": {
+     "tp": 149,
+     "fp": 40,
+     "fn": 20,
+     "precision": 0.7883597883597884,
+     "recall": 0.8816568047337278,
+     "f1": 0.8324022346368716,
      "gt_count": 169,
+     "pred_count": 189
     },
     "Nostril External of Nucleus": {
+     "tp": 189,
+     "fp": 73,
+     "fn": 18,
+     "precision": 0.7213740458015268,
+     "recall": 0.9130434782608695,
+     "f1": 0.8059701492537313,
      "gt_count": 207,
+     "pred_count": 262
     }
    },
    "n_gt": 376,
+   "n_pred": 451
   },
   "counts": {
    "Cell, Antibody +": {
     "gt": 61,
+    "pred": 75,
+    "abs_err": 14,
+    "rel_err": 0.22950819672131148
    },
    "Cell, Antibody -": {
     "gt": 1,
+    "pred": 2,
+    "abs_err": 1,
+    "rel_err": 1.0
    },
    "Cell, Nostril, Antibody +": {
     "gt": 122,
+    "pred": 125,
+    "abs_err": 3,
+    "rel_err": 0.02459016393442623
    },
    "Cell, Nostril, Antibody -": {
     "gt": 0,
    },
    "Nostril in Nucleus": {
     "gt": 169,
+    "pred": 189,
+    "abs_err": 20,
+    "rel_err": 0.11834319526627218
    },
    "Nostril External of Nucleus": {
     "gt": 207,
+    "pred": 262,
+    "abs_err": 55,
+    "rel_err": 0.26570048309178745
    }
   },
   "fp_analysis": null,
   "pairs": null,
   "summary": {
+   "cell_det_f1": 0.8704663212435233,
+   "cell_det_precision": 0.8316831683168316,
+   "cell_det_recall": 0.9130434782608695,
+   "cell_mean_iou": 0.7832369871445182,
+   "cell_det_f1_iou30": 0.922279792746114,
+   "cell_det_f1_centroid": 0.9326424870466321,
+   "cell_attr_antibody": 0.9940476190476191,
+   "cell_attr_nostril": 0.9702380952380952,
+   "cell_attr_both": 0.9642857142857143,
+   "nos_det_f1": 0.8391777509068925,
+   "nos_det_precision": 0.7694013303769401,
    "nos_det_recall": 0.9228723404255319,
+   "nos_type_acc": 0.9740634005763689,
+   "f1[Cell, Antibody +]": 0.7647058823529411,
+   "f1[Cell, Antibody -]": 0.6666666666666666,
+   "f1[Cell, Nostril, Antibody +]": 0.8825910931174089,
    "f1[Cell, Nostril, Antibody -]": 0.0,
+   "f1[Nostril in Nucleus]": 0.8324022346368716,
+   "f1[Nostril External of Nucleus]": 0.8059701492537313
   }
  },
  "C112": {
   "cells": {
    "detection": {
+    "tp": 130,
+    "fp": 49,
+    "fn": 16,
+    "precision": 0.7262569832402235,
+    "recall": 0.8904109589041096,
+    "f1": 0.8,
+    "mean_iou_matched": 0.7378347860378052,
     "iou30": {
+     "tp": 141,
+     "fp": 38,
+     "fn": 5,
+     "precision": 0.7877094972067039,
+     "recall": 0.9657534246575342,
+     "f1": 0.8676923076923078
     },
     "centroid": {
+     "tp": 143,
+     "fp": 36,
+     "fn": 3,
+     "precision": 0.7988826815642458,
+     "recall": 0.9794520547945206,
+     "f1": 0.88
     }
    },
    "attribute_accuracy": {
+    "antibody": 1.0,
+    "nostril": 0.9923076923076923,
+    "both": 0.9923076923076923,
+    "n_matched": 130
    },
    "per_class": {
     "Cell, Antibody +": {
+     "tp": 109,
+     "fp": 43,
+     "fn": 15,
+     "precision": 0.7171052631578947,
+     "recall": 0.8790322580645161,
+     "f1": 0.7898550724637682,
      "gt_count": 124,
+     "pred_count": 152
     },
     "Cell, Antibody -": {
+     "tp": 5,
      "fp": 1,
+     "fn": 0,
+     "precision": 0.8333333333333334,
+     "recall": 1.0,
+     "f1": 0.9090909090909091,
      "gt_count": 5,
+     "pred_count": 6
     },
     "Cell, Nostril, Antibody +": {
      "tp": 15,
+     "fp": 6,
      "fn": 2,
+     "precision": 0.7142857142857143,
      "recall": 0.8823529411764706,
+     "f1": 0.7894736842105262,
      "gt_count": 17,
+     "pred_count": 21
     },
     "Cell, Nostril, Antibody -": {
      "tp": 0,
    },
    "confusion": {
     "Cell, Antibody +": {
+     "Cell, Antibody +": 109,
      "Cell, Antibody -": 0,
+     "Cell, Nostril, Antibody +": 1,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Antibody -": {
+     "Cell, Antibody +": 0,
+     "Cell, Antibody -": 5,
      "Cell, Nostril, Antibody +": 0,
      "Cell, Nostril, Antibody -": 0
     },
     }
    },
    "n_gt": 146,
+   "n_pred": 179
   },
   "nostrils": {
    "detection": {
     "tp": 25,
+    "fp": 11,
     "fn": 0,
+    "precision": 0.6944444444444444,
     "recall": 1.0,
+    "f1": 0.819672131147541,
+    "mean_dist_matched": 1.0636624208995131
    },
    "type_accuracy": 1.0,
    "per_class": {
     "Nostril in Nucleus": {
      "tp": 23,
+     "fp": 4,
      "fn": 0,
+     "precision": 0.8518518518518519,
      "recall": 1.0,
+     "f1": 0.92,
      "gt_count": 23,
+     "pred_count": 27
     },
     "Nostril External of Nucleus": {
      "tp": 2,
+     "fp": 7,
      "fn": 0,
+     "precision": 0.2222222222222222,
      "recall": 1.0,
+     "f1": 0.3636363636363636,
      "gt_count": 2,
+     "pred_count": 9
     }
    },
    "n_gt": 25,
+   "n_pred": 36
   },
   "counts": {
    "Cell, Antibody +": {
     "gt": 124,
+    "pred": 152,
+    "abs_err": 28,
+    "rel_err": 0.22580645161290322
    },
    "Cell, Antibody -": {
     "gt": 5,
+    "pred": 6,
+    "abs_err": 1,
+    "rel_err": 0.2
    },
    "Cell, Nostril, Antibody +": {
     "gt": 17,
+    "pred": 21,
+    "abs_err": 4,
+    "rel_err": 0.23529411764705882
    },
    "Cell, Nostril, Antibody -": {
     "gt": 0,
    },
    "Nostril in Nucleus": {
     "gt": 23,
+    "pred": 27,
+    "abs_err": 4,
+    "rel_err": 0.17391304347826086
    },
    "Nostril External of Nucleus": {
     "gt": 2,
+    "pred": 9,
+    "abs_err": 7,
+    "rel_err": 3.5
    }
   },
   "fp_analysis": null,
   "pairs": null,
   "summary": {
+   "cell_det_f1": 0.8,
+   "cell_det_precision": 0.7262569832402235,
+   "cell_det_recall": 0.8904109589041096,
+   "cell_mean_iou": 0.7378347860378052,
+   "cell_det_f1_iou30": 0.8676923076923078,
+   "cell_det_f1_centroid": 0.88,
+   "cell_attr_antibody": 1.0,
+   "cell_attr_nostril": 0.9923076923076923,
+   "cell_attr_both": 0.9923076923076923,
+   "nos_det_f1": 0.819672131147541,
+   "nos_det_precision": 0.6944444444444444,
    "nos_det_recall": 1.0,
    "nos_type_acc": 1.0,
+   "f1[Cell, Antibody +]": 0.7898550724637682,
+   "f1[Cell, Antibody -]": 0.9090909090909091,
+   "f1[Cell, Nostril, Antibody +]": 0.7894736842105262,
    "f1[Cell, Nostril, Antibody -]": 0.0,
+   "f1[Nostril in Nucleus]": 0.92,
+   "f1[Nostril External of Nucleus]": 0.3636363636363636
   }
  },
  "F100": {
   "cells": {
    "detection": {
+    "tp": 172,
+    "fp": 30,
+    "fn": 14,
+    "precision": 0.8514851485148515,
+    "recall": 0.9247311827956989,
+    "f1": 0.8865979381443299,
+    "mean_iou_matched": 0.7905866698793166,
     "iou30": {
+     "tp": 182,
+     "fp": 20,
+     "fn": 4,
+     "precision": 0.900990099009901,
+     "recall": 0.978494623655914,
+     "f1": 0.9381443298969072
     },
     "centroid": {
+     "tp": 180,
+     "fp": 22,
+     "fn": 6,
+     "precision": 0.8910891089108911,
+     "recall": 0.967741935483871,
+     "f1": 0.9278350515463919
     }
    },
    "attribute_accuracy": {
+    "antibody": 0.9941860465116279,
+    "nostril": 0.9534883720930233,
+    "both": 0.9476744186046512,
+    "n_matched": 172
    },
    "per_class": {
     "Cell, Antibody +": {
+     "tp": 80,
+     "fp": 26,
+     "fn": 7,
+     "precision": 0.7547169811320755,
+     "recall": 0.9195402298850575,
+     "f1": 0.8290155440414508,
      "gt_count": 87,
+     "pred_count": 106
     },
     "Cell, Antibody -": {
+     "tp": 5,
+     "fp": 6,
+     "fn": 1,
+     "precision": 0.45454545454545453,
+     "recall": 0.8333333333333334,
+     "f1": 0.5882352941176471,
      "gt_count": 6,
+     "pred_count": 11
     },
     "Cell, Nostril, Antibody +": {
+     "tp": 78,
+     "fp": 7,
+     "fn": 15,
+     "precision": 0.9176470588235294,
+     "recall": 0.8387096774193549,
+     "f1": 0.8764044943820225,
      "gt_count": 93,
+     "pred_count": 85
     },
     "Cell, Nostril, Antibody -": {
      "tp": 0,
    },
    "confusion": {
     "Cell, Antibody +": {
+     "Cell, Antibody +": 80,
      "Cell, Antibody -": 1,
+     "Cell, Nostril, Antibody +": 1,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Antibody -": {
+     "Cell, Antibody +": 0,
+     "Cell, Antibody -": 5,
      "Cell, Nostril, Antibody +": 0,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody +": {
+     "Cell, Antibody +": 7,
      "Cell, Antibody -": 0,
+     "Cell, Nostril, Antibody +": 78,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody -": {
     }
    },
    "n_gt": 186,
+   "n_pred": 202
   },
   "nostrils": {
    "detection": {
+    "tp": 144,
+    "fp": 13,
+    "fn": 18,
+    "precision": 0.9171974522292994,
+    "recall": 0.8888888888888888,
+    "f1": 0.90282131661442,
+    "mean_dist_matched": 1.0960012345310384
+   },
+   "type_accuracy": 0.9930555555555556,
    "per_class": {
     "Nostril in Nucleus": {
+     "tp": 103,
+     "fp": 13,
+     "fn": 15,
+     "precision": 0.8879310344827587,
+     "recall": 0.8728813559322034,
+     "f1": 0.8803418803418803,
      "gt_count": 118,
+     "pred_count": 116
     },
     "Nostril External of Nucleus": {
+     "tp": 40,
+     "fp": 1,
+     "fn": 4,
+     "precision": 0.975609756097561,
+     "recall": 0.9090909090909091,
+     "f1": 0.9411764705882352,
      "gt_count": 44,
      "pred_count": 41
     }
    },
    "n_gt": 162,
+   "n_pred": 157
   },
   "counts": {
    "Cell, Antibody +": {
     "gt": 87,
+    "pred": 106,
+    "abs_err": 19,
+    "rel_err": 0.21839080459770116
    },
    "Cell, Antibody -": {
     "gt": 6,
+    "pred": 11,
+    "abs_err": 5,
+    "rel_err": 0.8333333333333334
    },
    "Cell, Nostril, Antibody +": {
     "gt": 93,
+    "pred": 85,
+    "abs_err": 8,
+    "rel_err": 0.08602150537634409
    },
    "Cell, Nostril, Antibody -": {
     "gt": 0,
    },
    "Nostril in Nucleus": {
     "gt": 118,
+    "pred": 116,
+    "abs_err": 2,
+    "rel_err": 0.01694915254237288
    },
    "Nostril External of Nucleus": {
     "gt": 44,
   "fp_analysis": null,
   "pairs": null,
   "summary": {
+   "cell_det_f1": 0.8865979381443299,
+   "cell_det_precision": 0.8514851485148515,
+   "cell_det_recall": 0.9247311827956989,
+   "cell_mean_iou": 0.7905866698793166,
+   "cell_det_f1_iou30": 0.9381443298969072,
+   "cell_det_f1_centroid": 0.9278350515463919,
+   "cell_attr_antibody": 0.9941860465116279,
+   "cell_attr_nostril": 0.9534883720930233,
+   "cell_attr_both": 0.9476744186046512,
+   "nos_det_f1": 0.90282131661442,
+   "nos_det_precision": 0.9171974522292994,
+   "nos_det_recall": 0.8888888888888888,
+   "nos_type_acc": 0.9930555555555556,
+   "f1[Cell, Antibody +]": 0.8290155440414508,
+   "f1[Cell, Antibody -]": 0.5882352941176471,
+   "f1[Cell, Nostril, Antibody +]": 0.8764044943820225,
    "f1[Cell, Nostril, Antibody -]": 0.0,
+   "f1[Nostril in Nucleus]": 0.8803418803418803,
+   "f1[Nostril External of Nucleus]": 0.9411764705882352
   }
  },
  "F110": {
   "cells": {
    "detection": {
+    "tp": 122,
+    "fp": 22,
+    "fn": 6,
+    "precision": 0.8472222222222222,
+    "recall": 0.953125,
+    "f1": 0.8970588235294118,
+    "mean_iou_matched": 0.7899681957823826,
     "iou30": {
+     "tp": 125,
+     "fp": 19,
+     "fn": 3,
+     "precision": 0.8680555555555556,
+     "recall": 0.9765625,
+     "f1": 0.9191176470588235
     },
     "centroid": {
+     "tp": 126,
+     "fp": 18,
+     "fn": 2,
+     "precision": 0.875,
+     "recall": 0.984375,
+     "f1": 0.9264705882352942
     }
    },
    "attribute_accuracy": {
+    "antibody": 1.0,
+    "nostril": 0.9754098360655737,
+    "both": 0.9754098360655737,
+    "n_matched": 122
    },
    "per_class": {
     "Cell, Antibody +": {
+     "tp": 85,
+     "fp": 15,
+     "fn": 2,
+     "precision": 0.85,
+     "recall": 0.9770114942528736,
+     "f1": 0.9090909090909092,
      "gt_count": 87,
+     "pred_count": 100
     },
     "Cell, Antibody -": {
+     "tp": 6,
+     "fp": 5,
+     "fn": 1,
+     "precision": 0.5454545454545454,
+     "recall": 0.8571428571428571,
+     "f1": 0.6666666666666665,
      "gt_count": 7,
+     "pred_count": 11
     },
     "Cell, Nostril, Antibody +": {
+     "tp": 28,
      "fp": 5,
+     "fn": 5,
+     "precision": 0.8484848484848485,
+     "recall": 0.8484848484848485,
+     "f1": 0.8484848484848486,
      "gt_count": 33,
+     "pred_count": 33
     },
     "Cell, Nostril, Antibody -": {
      "tp": 0,
    },
    "confusion": {
     "Cell, Antibody +": {
+     "Cell, Antibody +": 85,
      "Cell, Antibody -": 0,
      "Cell, Nostril, Antibody +": 0,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Antibody -": {
+     "Cell, Antibody +": 0,
+     "Cell, Antibody -": 6,
      "Cell, Nostril, Antibody +": 0,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody +": {
+     "Cell, Antibody +": 3,
      "Cell, Antibody -": 0,
+     "Cell, Nostril, Antibody +": 28,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody -": {
     }
    },
    "n_gt": 128,
+   "n_pred": 144
   },
   "nostrils": {
    "detection": {
+    "tp": 164,
+    "fp": 12,
+    "fn": 30,
+    "precision": 0.9318181818181818,
+    "recall": 0.845360824742268,
+    "f1": 0.8864864864864865,
+    "mean_dist_matched": 0.8436248674412936
+   },
+   "type_accuracy": 0.9878048780487805,
    "per_class": {
     "Nostril in Nucleus": {
+     "tp": 46,
+     "fp": 5,
+     "fn": 11,
+     "precision": 0.9019607843137255,
+     "recall": 0.8070175438596491,
+     "f1": 0.8518518518518519,
      "gt_count": 57,
+     "pred_count": 51
     },
     "Nostril External of Nucleus": {
+     "tp": 116,
+     "fp": 9,
+     "fn": 21,
+     "precision": 0.928,
+     "recall": 0.8467153284671532,
+     "f1": 0.8854961832061069,
      "gt_count": 137,
+     "pred_count": 125
     }
    },
    "n_gt": 194,
+   "n_pred": 176
   },
   "counts": {
    "Cell, Antibody +": {
     "gt": 87,
+    "pred": 100,
+    "abs_err": 13,
+    "rel_err": 0.14942528735632185
    },
    "Cell, Antibody -": {
     "gt": 7,
+    "pred": 11,
+    "abs_err": 4,
+    "rel_err": 0.5714285714285714
    },
    "Cell, Nostril, Antibody +": {
     "gt": 33,
+    "pred": 33,
+    "abs_err": 0,
+    "rel_err": 0.0
    },
    "Cell, Nostril, Antibody -": {
     "gt": 1,
    },
    "Nostril in Nucleus": {
     "gt": 57,
+    "pred": 51,
+    "abs_err": 6,
+    "rel_err": 0.10526315789473684
    },
    "Nostril External of Nucleus": {
     "gt": 137,
+    "pred": 125,
+    "abs_err": 12,
+    "rel_err": 0.08759124087591241
    }
   },
   "fp_analysis": null,
   "pairs": null,
   "summary": {
+   "cell_det_f1": 0.8970588235294118,
+   "cell_det_precision": 0.8472222222222222,
+   "cell_det_recall": 0.953125,
+   "cell_mean_iou": 0.7899681957823826,
+   "cell_det_f1_iou30": 0.9191176470588235,
+   "cell_det_f1_centroid": 0.9264705882352942,
+   "cell_attr_antibody": 1.0,
+   "cell_attr_nostril": 0.9754098360655737,
+   "cell_attr_both": 0.9754098360655737,
+   "nos_det_f1": 0.8864864864864865,
+   "nos_det_precision": 0.9318181818181818,
+   "nos_det_recall": 0.845360824742268,
+   "nos_type_acc": 0.9878048780487805,
+   "f1[Cell, Antibody +]": 0.9090909090909092,
+   "f1[Cell, Antibody -]": 0.6666666666666665,
+   "f1[Cell, Nostril, Antibody +]": 0.8484848484848486,
    "f1[Cell, Nostril, Antibody -]": 0.0,
+   "f1[Nostril in Nucleus]": 0.8518518518518519,
+   "f1[Nostril External of Nucleus]": 0.8854961832061069
   }
  },
  "F113": {
   "cells": {
    "detection": {
+    "tp": 140,
+    "fp": 27,
+    "fn": 23,
+    "precision": 0.8383233532934131,
+    "recall": 0.8588957055214724,
+    "f1": 0.8484848484848485,
+    "mean_iou_matched": 0.7555449040647428,
     "iou30": {
+     "tp": 155,
+     "fp": 12,
+     "fn": 8,
+     "precision": 0.9281437125748503,
+     "recall": 0.950920245398773,
+     "f1": 0.9393939393939393
     },
     "centroid": {
+     "tp": 153,
+     "fp": 14,
+     "fn": 10,
+     "precision": 0.9161676646706587,
+     "recall": 0.9386503067484663,
+     "f1": 0.9272727272727274
     }
    },
    "attribute_accuracy": {
     "antibody": 1.0,
+    "nostril": 0.9785714285714285,
+    "both": 0.9785714285714285,
+    "n_matched": 140
    },
    "per_class": {
     "Cell, Antibody +": {
+     "tp": 117,
+     "fp": 26,
+     "fn": 21,
+     "precision": 0.8181818181818182,
+     "recall": 0.8478260869565217,
+     "f1": 0.8327402135231317,
      "gt_count": 138,
+     "pred_count": 143
     },
     "Cell, Antibody -": {
      "tp": 2,
+     "fp": 1,
      "fn": 1,
+     "precision": 0.6666666666666666,
      "recall": 0.6666666666666666,
+     "f1": 0.6666666666666666,
      "gt_count": 3,
+     "pred_count": 3
     },
     "Cell, Nostril, Antibody +": {
      "tp": 18,
+     "fp": 3,
      "fn": 4,
+     "precision": 0.8571428571428571,
      "recall": 0.8181818181818182,
+     "f1": 0.8372093023255814,
      "gt_count": 22,
+     "pred_count": 21
     },
     "Cell, Nostril, Antibody -": {
      "tp": 0,
    },
    "confusion": {
     "Cell, Antibody +": {
+     "Cell, Antibody +": 117,
      "Cell, Antibody -": 0,
      "Cell, Nostril, Antibody +": 0,
      "Cell, Nostril, Antibody -": 0
     }
    },
    "n_gt": 163,
+   "n_pred": 167
   },
   "nostrils": {
    "detection": {
+    "tp": 29,
+    "fp": 0,
+    "fn": 5,
+    "precision": 1.0,
+    "recall": 0.8529411764705882,
+    "f1": 0.9206349206349206,
+    "mean_dist_matched": 0.9723494301213084
+   },
+   "type_accuracy": 0.9310344827586207,
    "per_class": {
     "Nostril in Nucleus": {
+     "tp": 22,
+     "fp": 1,
+     "fn": 4,
+     "precision": 0.9565217391304348,
+     "recall": 0.8461538461538461,
+     "f1": 0.8979591836734695,
      "gt_count": 26,
      "pred_count": 23
     },
     "Nostril External of Nucleus": {
+     "tp": 5,
      "fp": 1,
+     "fn": 3,
+     "precision": 0.8333333333333334,
+     "recall": 0.625,
+     "f1": 0.7142857142857143,
      "gt_count": 8,
+     "pred_count": 6
     }
    },
    "n_gt": 34,
+   "n_pred": 29
   },
   "counts": {
    "Cell, Antibody +": {
     "gt": 138,
+    "pred": 143,
+    "abs_err": 5,
+    "rel_err": 0.036231884057971016
    },
    "Cell, Antibody -": {
     "gt": 3,
+    "pred": 3,
+    "abs_err": 0,
+    "rel_err": 0.0
    },
    "Cell, Nostril, Antibody +": {
     "gt": 22,
+    "pred": 21,
+    "abs_err": 1,
+    "rel_err": 0.045454545454545456
    },
    "Cell, Nostril, Antibody -": {
     "gt": 0,
    },
    "Nostril External of Nucleus": {
     "gt": 8,
+    "pred": 6,
+    "abs_err": 2,
+    "rel_err": 0.25
    }
   },
   "fp_analysis": null,
   "pairs": null,
   "summary": {
+   "cell_det_f1": 0.8484848484848485,
+   "cell_det_precision": 0.8383233532934131,
+   "cell_det_recall": 0.8588957055214724,
+   "cell_mean_iou": 0.7555449040647428,
+   "cell_det_f1_iou30": 0.9393939393939393,
+   "cell_det_f1_centroid": 0.9272727272727274,
    "cell_attr_antibody": 1.0,
+   "cell_attr_nostril": 0.9785714285714285,
+   "cell_attr_both": 0.9785714285714285,
+   "nos_det_f1": 0.9206349206349206,
+   "nos_det_precision": 1.0,
+   "nos_det_recall": 0.8529411764705882,
+   "nos_type_acc": 0.9310344827586207,
+   "f1[Cell, Antibody +]": 0.8327402135231317,
+   "f1[Cell, Antibody -]": 0.6666666666666666,
+   "f1[Cell, Nostril, Antibody +]": 0.8372093023255814,
    "f1[Cell, Nostril, Antibody -]": 0.0,
+   "f1[Nostril in Nucleus]": 0.8979591836734695,
+   "f1[Nostril External of Nucleus]": 0.7142857142857143
   }
  },
  "F114": {
   "cells": {
    "detection": {
+    "tp": 226,
+    "fp": 30,
+    "fn": 16,
+    "precision": 0.8828125,
+    "recall": 0.9338842975206612,
+    "f1": 0.9076305220883534,
+    "mean_iou_matched": 0.8136301605298976,
     "iou30": {
+     "tp": 236,
+     "fp": 20,
+     "fn": 6,
+     "precision": 0.921875,
+     "recall": 0.9752066115702479,
+     "f1": 0.9477911646586346
     },
     "centroid": {
+     "tp": 236,
+     "fp": 20,
+     "fn": 6,
+     "precision": 0.921875,
+     "recall": 0.9752066115702479,
+     "f1": 0.9477911646586346
     }
    },
    "attribute_accuracy": {
+    "antibody": 0.9911504424778761,
+    "nostril": 0.9601769911504425,
+    "both": 0.9513274336283186,
+    "n_matched": 226
    },
    "per_class": {
     "Cell, Antibody +": {
+     "tp": 157,
+     "fp": 20,
+     "fn": 18,
+     "precision": 0.8870056497175142,
+     "recall": 0.8971428571428571,
+     "f1": 0.8920454545454545,
      "gt_count": 175,
+     "pred_count": 177
     },
     "Cell, Antibody -": {
+     "tp": 21,
+     "fp": 9,
+     "fn": 0,
+     "precision": 0.7,
+     "recall": 1.0,
+     "f1": 0.8235294117647058,
      "gt_count": 21,
+     "pred_count": 30
     },
     "Cell, Nostril, Antibody +": {
+     "tp": 36,
+     "fp": 12,
+     "fn": 6,
+     "precision": 0.75,
+     "recall": 0.8571428571428571,
+     "f1": 0.7999999999999999,
      "gt_count": 42,
+     "pred_count": 48
     },
     "Cell, Nostril, Antibody -": {
+     "tp": 1,
      "fp": 0,
+     "fn": 3,
+     "precision": 1.0,
+     "recall": 0.25,
+     "f1": 0.4,
      "gt_count": 4,
+     "pred_count": 1
     }
    },
    "confusion": {
     "Cell, Antibody +": {
+     "Cell, Antibody +": 157,
+     "Cell, Antibody -": 2,
+     "Cell, Nostril, Antibody +": 5,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Antibody -": {
+     "Cell, Antibody +": 0,
+     "Cell, Antibody -": 21,
      "Cell, Nostril, Antibody +": 0,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody +": {
+     "Cell, Antibody +": 1,
      "Cell, Antibody -": 0,
+     "Cell, Nostril, Antibody +": 36,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody -": {
      "Cell, Antibody +": 0,
+     "Cell, Antibody -": 3,
      "Cell, Nostril, Antibody +": 0,
+     "Cell, Nostril, Antibody -": 1
     }
    },
    "n_gt": 242,
+   "n_pred": 256
   },
   "nostrils": {
    "detection": {
+    "tp": 85,
+    "fp": 31,
+    "fn": 5,
+    "precision": 0.7327586206896551,
+    "recall": 0.9444444444444444,
+    "f1": 0.8252427184466019,
+    "mean_dist_matched": 1.0045528563892454
    },
    "type_accuracy": 1.0,
    "per_class": {
     "Nostril in Nucleus": {
+     "tp": 48,
+     "fp": 11,
+     "fn": 4,
+     "precision": 0.8135593220338984,
+     "recall": 0.9230769230769231,
+     "f1": 0.8648648648648649,
      "gt_count": 52,
+     "pred_count": 59
     },
     "Nostril External of Nucleus": {
+     "tp": 37,
+     "fp": 20,
+     "fn": 1,
+     "precision": 0.6491228070175439,
+     "recall": 0.9736842105263158,
+     "f1": 0.7789473684210527,
      "gt_count": 38,
+     "pred_count": 57
     }
    },
    "n_gt": 90,
+   "n_pred": 116
   },
   "counts": {
    "Cell, Antibody +": {
     "gt": 175,
+    "pred": 177,
+    "abs_err": 2,
+    "rel_err": 0.011428571428571429
    },
    "Cell, Antibody -": {
     "gt": 21,
+    "pred": 30,
+    "abs_err": 9,
+    "rel_err": 0.42857142857142855
    },
    "Cell, Nostril, Antibody +": {
     "gt": 42,
+    "pred": 48,
+    "abs_err": 6,
+    "rel_err": 0.14285714285714285
    },
    "Cell, Nostril, Antibody -": {
     "gt": 4,
+    "pred": 1,
+    "abs_err": 3,
+    "rel_err": 0.75
    },
    "Nostril in Nucleus": {
     "gt": 52,
+    "pred": 59,
+    "abs_err": 7,
+    "rel_err": 0.1346153846153846
    },
    "Nostril External of Nucleus": {
     "gt": 38,
+    "pred": 57,
+    "abs_err": 19,
+    "rel_err": 0.5
    }
   },
   "fp_analysis": null,
   "pairs": null,
   "summary": {
+   "cell_det_f1": 0.9076305220883534,
+   "cell_det_precision": 0.8828125,
+   "cell_det_recall": 0.9338842975206612,
+   "cell_mean_iou": 0.8136301605298976,
+   "cell_det_f1_iou30": 0.9477911646586346,
+   "cell_det_f1_centroid": 0.9477911646586346,
+   "cell_attr_antibody": 0.9911504424778761,
+   "cell_attr_nostril": 0.9601769911504425,
+   "cell_attr_both": 0.9513274336283186,
+   "nos_det_f1": 0.8252427184466019,
+   "nos_det_precision": 0.7327586206896551,
+   "nos_det_recall": 0.9444444444444444,
    "nos_type_acc": 1.0,
+   "f1[Cell, Antibody +]": 0.8920454545454545,
+   "f1[Cell, Antibody -]": 0.8235294117647058,
+   "f1[Cell, Nostril, Antibody +]": 0.7999999999999999,
+   "f1[Cell, Nostril, Antibody -]": 0.4,
+   "f1[Nostril in Nucleus]": 0.8648648648648649,
+   "f1[Nostril External of Nucleus]": 0.7789473684210527
   }
  },
  "F99": {
   "cells": {
    "detection": {
+    "tp": 241,
+    "fp": 35,
+    "fn": 10,
+    "precision": 0.8731884057971014,
+    "recall": 0.9601593625498008,
+    "f1": 0.9146110056925996,
+    "mean_iou_matched": 0.8180266624759959,
     "iou30": {
+     "tp": 247,
+     "fp": 29,
+     "fn": 4,
+     "precision": 0.894927536231884,
+     "recall": 0.9840637450199203,
+     "f1": 0.937381404174573
     },
     "centroid": {
+     "tp": 246,
+     "fp": 30,
+     "fn": 5,
+     "precision": 0.8913043478260869,
+     "recall": 0.9800796812749004,
+     "f1": 0.9335863377609108
     }
    },
    "attribute_accuracy": {
+    "antibody": 0.979253112033195,
+    "nostril": 0.991701244813278,
+    "both": 0.970954356846473,
+    "n_matched": 241
    },
    "per_class": {
     "Cell, Antibody +": {
+     "tp": 173,
+     "fp": 26,
+     "fn": 10,
+     "precision": 0.8693467336683417,
+     "recall": 0.9453551912568307,
+     "f1": 0.9057591623036649,
      "gt_count": 183,
+     "pred_count": 199
     },
     "Cell, Antibody -": {
+     "tp": 57,
+     "fp": 16,
+     "fn": 5,
+     "precision": 0.7808219178082192,
+     "recall": 0.9193548387096774,
+     "f1": 0.8444444444444444,
      "gt_count": 62,
+     "pred_count": 73
     },
     "Cell, Nostril, Antibody +": {
+     "tp": 4,
+     "fp": 0,
+     "fn": 2,
+     "precision": 1.0,
+     "recall": 0.6666666666666666,
+     "f1": 0.8,
      "gt_count": 6,
      "pred_count": 4
     },
    },
    "confusion": {
     "Cell, Antibody +": {
+     "Cell, Antibody +": 173,
      "Cell, Antibody -": 4,
      "Cell, Nostril, Antibody +": 0,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Antibody -": {
+     "Cell, Antibody +": 1,
+     "Cell, Antibody -": 57,
      "Cell, Nostril, Antibody +": 0,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody +": {
      "Cell, Antibody +": 2,
      "Cell, Antibody -": 0,
+     "Cell, Nostril, Antibody +": 4,
      "Cell, Nostril, Antibody -": 0
     },
     "Cell, Nostril, Antibody -": {
     }
    },
    "n_gt": 251,
+   "n_pred": 276
   },
   "nostrils": {
    "detection": {
+    "tp": 20,
+    "fp": 2,
+    "fn": 2,
+    "precision": 0.9090909090909091,
+    "recall": 0.9090909090909091,
+    "f1": 0.9090909090909091,
+    "mean_dist_matched": 0.8019220998015626
    },
    "type_accuracy": 1.0,
    "per_class": {
      "pred_count": 4
     },
     "Nostril External of Nucleus": {
+     "tp": 16,
+     "fp": 2,
+     "fn": 0,
+     "precision": 0.8888888888888888,
+     "recall": 1.0,
+     "f1": 0.9411764705882353,
      "gt_count": 16,
+     "pred_count": 18
     }
    },
    "n_gt": 22,
+   "n_pred": 22
   },
   "counts": {
    "Cell, Antibody +": {
     "gt": 183,
+    "pred": 199,
+    "abs_err": 16,
+    "rel_err": 0.08743169398907104
    },
    "Cell, Antibody -": {
     "gt": 62,
+    "pred": 73,
+    "abs_err": 11,
+    "rel_err": 0.1774193548387097
    },
    "Cell, Nostril, Antibody +": {
     "gt": 6,
    },
    "Nostril External of Nucleus": {
     "gt": 16,
+    "pred": 18,
+    "abs_err": 2,
+    "rel_err": 0.125
    }
   },
   "fp_analysis": null,
   "pairs": null,
   "summary": {
+   "cell_det_f1": 0.9146110056925996,
+   "cell_det_precision": 0.8731884057971014,
+   "cell_det_recall": 0.9601593625498008,
+   "cell_mean_iou": 0.8180266624759959,
+   "cell_det_f1_iou30": 0.937381404174573,
+   "cell_det_f1_centroid": 0.9335863377609108,
+   "cell_attr_antibody": 0.979253112033195,
+   "cell_attr_nostril": 0.991701244813278,
+   "cell_attr_both": 0.970954356846473,
+   "nos_det_f1": 0.9090909090909091,
+   "nos_det_precision": 0.9090909090909091,
+   "nos_det_recall": 0.9090909090909091,
    "nos_type_acc": 1.0,
+   "f1[Cell, Antibody +]": 0.9057591623036649,
+   "f1[Cell, Antibody -]": 0.8444444444444444,
+   "f1[Cell, Nostril, Antibody +]": 0.8,
    "f1[Cell, Nostril, Antibody -]": 0.0,
    "f1[Nostril in Nucleus]": 0.8,
+   "f1[Nostril External of Nucleus]": 0.9411764705882353
   }
  }
 }

scripts/build_summary.py CHANGED Viewed

@@ -1,7 +1,11 @@
 #!/usr/bin/env python3
-"""Write reports/SUMMARY.md from runs/cv/cv_results.json and reports/final/final_fit.json."""
 from __future__ import annotations
 import json
 import os
 import sys
@@ -15,8 +19,14 @@ def pct(x: float) -> str:
     return f"{100 * x:.1f}%"
-def main():
-    cv = json.load(open(os.path.join(ROOT, "runs", "cv", "cv_results.json")))
     agg = cv["aggregate"]
     per = agg["per_image"]
     fit = json.load(open(os.path.join(ROOT, "reports", "final", "final_fit.json")))
@@ -32,6 +42,11 @@ def main():
          "(determinism), no-TTA / 4× / 8× test-time augmentation, 8 rotations and flips, 6 photometric perturbations, and 6 post-processing rule variants. "
          "The only choice made on held-out data is the rule that gives a nucleus the Nostril attribute: of three candidate rules, "
          "'contains an in-nucleus punctum' (the annotation protocol) scored best and is the default; see the rule-variant table in cv_report.md.", "",
          "## Headline numbers (micro-averaged over all held-out images, no TTA)", "",
          "| Measure | Precision | Recall | F1 |", "|---|---:|---:|---:|",
          f"| Nucleus detection, IoU ≥ 0.5 | {pct(det['cells']['precision'])} | {pct(det['cells']['recall'])} | {pct(det['cells']['f1'])} |",
@@ -57,21 +72,52 @@ def main():
         if "summary" in v:
             s = v["summary"]
             L.append(f"| {k} | {pct(s['cell_det_f1'])} | {pct(s['cell_attr_both'])} | {pct(s['nos_det_f1'])} |")
     sw = agg.get("antibody_threshold_sweep")
     if sw:
         best = max(sw["sweep"], key=lambda r: r["neg_f1"])
         L += ["", f"## Antibody − threshold (held-out matched nuclei: {sw['n_matched']}, of which {sw['n_negative']} negative)", "",
               f"Default threshold 0.5. Best Antibody − F1 in the sweep: {pct(best['neg_f1'])} at threshold {best['thr']:.2f} "
               f"(precision {pct(best['neg_precision'])}, recall {pct(best['neg_recall'])}); the full curve is in cv_report.md. "
-              "The default was kept because the class is dominated by a single image (F99 holds 62 of the 102 negatives)."]
     L += ["", "## What limits the scores", "",
-          "* **Touching nuclei are sometimes merged** into one outline (worst on F113); this is the main source of IoU-0.5 misses. The marker/threshold sweep (`runs/cv/postprocess_sweep.json`) is flat, so it is a network-border issue, not a post-processing one.",
-          "* **Annotation completeness**: most unmatched predictions overlap an annotated nucleus (outline disagreement) and the rest are dimmer, smaller objects the annotator skipped; many faint red dots were likewise not annotated, which caps external-punctum precision.",
-          "* **Antibody − is rare** (102 of 1569 nuclei, 62 in one image), and **Cell, Nostril, Antibody −** has five examples; their scores are not reliable.",
           "* Scores are from nine images of one experiment (35-day spinal cord, Iba1/TMEM119). New staining batches or magnifications need a check against a few annotated images.",
           "", "## Reproduce", "",
-          "```bash", "python evaluate.py cv --data-dir <folder with TIFs and CVAT zips> --runs-dir runs/cv --iters 3000",
-          "python evaluate.py final --data-dir <folder> --weights weights/cellquant.pt", "python scripts/build_summary.py", "```", ""]
     out = os.path.join(ROOT, "reports", "SUMMARY.md")
     with open(out, "w") as f:
         f.write("\n".join(L))

 #!/usr/bin/env python3
+"""Write reports/SUMMARY.md from a cross-validation run and reports/final/final_fit.json.
+    python scripts/build_summary.py [--cv-dir runs/cv_v2] [--recipe v2]
+"""
 from __future__ import annotations
+import argparse
 import json
 import os
 import sys
     return f"{100 * x:.1f}%"
+def main(argv=None):
+    ap = argparse.ArgumentParser()
+    ap.add_argument("--cv-dir", default="runs/cv", help="folder holding cv_results.json of the shipped recipe")
+    ap.add_argument("--compare", default="reports/cv/cv_results.json",
+                    help="cv_results.json of the previous recipe for the change table ('' = none)")
+    a = ap.parse_args(argv)
+    cv = json.load(open(os.path.join(ROOT, a.cv_dir, "cv_results.json")))
+    prev = json.load(open(os.path.join(ROOT, a.compare)))["aggregate"] if a.compare and os.path.exists(os.path.join(ROOT, a.compare)) else None
     agg = cv["aggregate"]
     per = agg["per_image"]
     fit = json.load(open(os.path.join(ROOT, "reports", "final", "final_fit.json")))
          "(determinism), no-TTA / 4× / 8× test-time augmentation, 8 rotations and flips, 6 photometric perturbations, and 6 post-processing rule variants. "
          "The only choice made on held-out data is the rule that gives a nucleus the Nostril attribute: of three candidate rules, "
          "'contains an in-nucleus punctum' (the annotation protocol) scored best and is the default; see the rule-variant table in cv_report.md.", "",
+         "Recipe (v2, shipped): ResNet34-UNet with a full-resolution stem; targets are a 3-px inner border ring (class weight 3), "
+         "a per-nucleus normalised distance-to-outline map that the watershed uses as its landscape (markers = distance > 0.5), "
+         "an antibody head whose loss up-weights antibody-negative pixels by sqrt(positive / negative pixels) (4.5 on the full set), "
+         "and 25 % of training crops centred on an antibody-negative nucleus. v1 (reports/cv/) used a 2-px ring, the interior map as "
+         "watershed landscape and an unweighted antibody loss; the change table below compares the two under the same protocol.", "",
          "## Headline numbers (micro-averaged over all held-out images, no TTA)", "",
          "| Measure | Precision | Recall | F1 |", "|---|---:|---:|---:|",
          f"| Nucleus detection, IoU ≥ 0.5 | {pct(det['cells']['precision'])} | {pct(det['cells']['recall'])} | {pct(det['cells']['f1'])} |",
         if "summary" in v:
             s = v["summary"]
             L.append(f"| {k} | {pct(s['cell_det_f1'])} | {pct(s['cell_attr_both'])} | {pct(s['nos_det_f1'])} |")
+    if prev:
+        pd, pd8 = prev["detection_base"], prev["detection_tta8"]
+        L += ["", "## Change from v1 (same protocol, same nine held-out images; `reports/cv_v2/comparison.md` has the full table)", "",
+              "| Measure | v1 | v2 | Δ (points) |", "|---|---:|---:|---:|"]
+        def row(name, x, y):
+            L.append(f"| {name} | {pct(x)} | {pct(y)} | {100 * (y - x):+.1f} |")
+        row("Nucleus F1, IoU ≥ 0.5", pd["cells"]["f1"], det["cells"]["f1"])
+        row("Nucleus precision, IoU ≥ 0.5", pd["cells"]["precision"], det["cells"]["precision"])
+        row("Nucleus recall, IoU ≥ 0.5", pd["cells"]["recall"], det["cells"]["recall"])
+        row("Nucleus F1, centroid rule", pd["cells_centroid"]["f1"], det["cells_centroid"]["f1"])
+        row("Nucleus F1, IoU ≥ 0.5, 8× TTA", pd8["cells"]["f1"], det8["cells"]["f1"])
+        row("Puncta F1", pd["nostrils"]["f1"], det["nostrils"]["f1"])
+        for lab in ALL_LABELS:
+            row(f"F1 [{lab}]", prev["micro_base"][lab]["f1"], agg["micro_base"][lab]["f1"])
+        pr, cr = prev["rule_variants"]["rule_nos=head_cell=contains"], agg["rule_variants"]["rule_nos=head_cell=contains"]
+        row("Attribute accuracy, both (each model's own matched nuclei)", pr["pooled_cell_attr_both"], cr["pooled_cell_attr_both"])
+        row("Punctum type accuracy (matched puncta)", pr["pooled_nos_type_acc"], cr["pooled_nos_type_acc"])
+        for k in sorted(per):
+            if k in prev["per_image"]:
+                row(f"Nuclei F1, {k} held out", prev["per_image"][k]["base"]["cell_det_f1"], per[k]["base"]["cell_det_f1"])
+        n_pred_v1 = pd["cells"]["tp"] + pd["cells"]["fp"]
+        n_pred_v2 = det["cells"]["tp"] + det["cells"]["fp"]
+        n_gt = det["cells"]["tp"] + det["cells"]["fn"]
+        L += ["", f"v2 misses far fewer nuclei ({pd['cells']['fn']} → {det['cells']['fn']} unmatched annotated nuclei, mostly touching nuclei that v1 "
+              f"merged) at the cost of more unmatched predictions ({pd['cells']['fp']} → {det['cells']['fp']}): it predicts {n_pred_v2} nuclei "
+              f"for {n_gt} annotated ones (v1: {n_pred_v1}). The extra predictions are dim objects the annotator skipped and outline/split "
+              "disagreements in roughly the proportion shown in the FP table of cv_report.md. On the nuclei matched by both models the "
+              "attribute accuracy differs by less than one point (see comparison.md), which is within the sampling error of ~1200 nuclei; "
+              "the antibody-negative class gains recall (26 % → 38 % on matched nuclei) and loses precision (74 % → 59 %)."]
     sw = agg.get("antibody_threshold_sweep")
     if sw:
         best = max(sw["sweep"], key=lambda r: r["neg_f1"])
         L += ["", f"## Antibody − threshold (held-out matched nuclei: {sw['n_matched']}, of which {sw['n_negative']} negative)", "",
               f"Default threshold 0.5. Best Antibody − F1 in the sweep: {pct(best['neg_f1'])} at threshold {best['thr']:.2f} "
               f"(precision {pct(best['neg_precision'])}, recall {pct(best['neg_recall'])}); the full curve is in cv_report.md. "
+              f"The default was kept because the class is dominated by a single image (F99 holds 62 of the {tot['Cell, Antibody -'] + tot['Cell, Nostril, Antibody -']} negatives)."]
     L += ["", "## What limits the scores", "",
+          "* **Touching nuclei** are still the main source of IoU-0.5 misses (worst on F113, the densest image). The distance-map watershed of v2 recovers many clusters that v1 merged (v1 marker/threshold sweeps in `reports/cv/postprocess_sweep.json` had shown post-processing alone could not), but some clusters are still merged and a few nuclei are now split in two.",
+          "* **Annotation completeness**: most unmatched predictions overlap an annotated nucleus (outline disagreement) and the rest are dimmer, smaller objects the annotator skipped; many faint red dots were likewise not annotated, which caps external-punctum precision. v2 detects more of these dim objects than v1, so it over-counts nuclei by about 13 % on the annotated images.",
+          f"* **Antibody − is rare** ({tot['Cell, Antibody -'] + tot['Cell, Nostril, Antibody -']} of {n_nuc} nuclei, 62 in one image), and **Cell, Nostril, Antibody −** has five examples; their scores are not reliable. The class-balanced loss of v2 trades antibody-negative precision for recall.",
           "* Scores are from nine images of one experiment (35-day spinal cord, Iba1/TMEM119). New staining batches or magnifications need a check against a few annotated images.",
           "", "## Reproduce", "",
+          "```bash", "python evaluate.py cv --data-dir <folder with TIFs and CVAT zips> --runs-dir runs/cv_v2 --iters 3000",
+          "python evaluate.py final --data-dir <folder> --weights weights/cellquant.pt",
+          "python scripts/compare_cv.py reports/cv/cv_results.json runs/cv_v2/cv_results.json --out reports/cv_v2/comparison.md",
+          "python scripts/build_summary.py --cv-dir runs/cv_v2", "```", ""]
     out = os.path.join(ROOT, "reports", "SUMMARY.md")
     with open(out, "w") as f:
         f.write("\n".join(L))

scripts/compare_attributes.py ADDED Viewed

	@@ -0,0 +1,113 @@

+#!/usr/bin/env python3
+"""Like-for-like attribute comparison of two cross-validation runs.
+    python scripts/compare_attributes.py --data-dir DATA --runs-a <folder with fold_*.pt> --runs-b <folder> [--device mps]
+Pooled attribute accuracy in cv_report.md is computed over the nuclei each model
+matched (IoU >= 0.5), so a model that matches more nuclei is scored on a harder
+set. This script predicts every held-out image with both fold checkpoints and
+reports the antibody / Nostril attribute accuracy on the nuclei matched by BOTH,
+plus the per-label count errors. It makes no choice; it only reads out numbers.
+"""
+from __future__ import annotations
+import argparse
+import json
+import os
+import sys
+import numpy as np
+sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+from cellquant.annotations import discover, load_image, rasterize  # noqa: E402
+from cellquant.infer import Predictor  # noqa: E402
+from cellquant.labels import ALL_LABELS, cell_attributes  # noqa: E402
+from cellquant.metrics import match_by_iou  # noqa: E402
+from cellquant.postprocess import PostConfig  # noqa: E402
+def main(argv=None):
+    ap = argparse.ArgumentParser()
+    ap.add_argument("--data-dir", required=True)
+    ap.add_argument("--runs-a", required=True)
+    ap.add_argument("--runs-b", required=True)
+    ap.add_argument("--names", default="v1,v2")
+    ap.add_argument("--device", default="auto")
+    ap.add_argument("--out", default="")
+    a = ap.parse_args(argv)
+    na, nb = a.names.split(",")
+    samples, _ = discover(a.data_dir, verbose=False)
+    tot = {na: {"ab": 0, "nos": 0, "both": 0}, nb: {"ab": 0, "nos": 0, "both": 0}, "n": 0}
+    neg = {na: {"tp": 0, "fp": 0, "fn": 0}, nb: {"tp": 0, "fp": 0, "fn": 0}}
+    counts = {lab: {"gt": 0, na: 0, nb: 0, f"abs_{na}": 0, f"abs_{nb}": 0} for lab in ALL_LABELS}
+    rows = []
+    for s in samples:
+        img = load_image(s.image_path)
+        R = rasterize(s.annotation, img.shape[0], img.shape[1])
+        preds = {}
+        for name, d in ((na, a.runs_a), (nb, a.runs_b)):
+            ck = os.path.join(d, f"fold_{s.key}.pt")
+            res, _ = Predictor(ck, device=a.device, tta=1).predict(img, PostConfig())
+            preds[name] = res
+            for lab in ALL_LABELS:
+                counts[lab][name] += res.counts[lab]
+                counts[lab][f"abs_{name}"] += abs(res.counts[lab] - s.annotation.count(lab))
+        for lab in ALL_LABELS:
+            counts[lab]["gt"] += s.annotation.count(lab)
+        m = {name: {g: p for g, p, _ in match_by_iou(R.cell_inst, preds[name].cell_inst, 0.5)} for name in preds}
+        common = sorted(set(m[na]) & set(m[nb]))
+        row = {"image": s.key, "n_common": len(common), f"n_{na}": len(m[na]), f"n_{nb}": len(m[nb])}
+        for name in (na, nb):
+            ab = ns = both = 0
+            for g in common:
+                ga, gn = cell_attributes(R.cell_labels[g])
+                pa, pn = cell_attributes(preds[name].cell_labels[m[name][g]])
+                ab += ga == pa
+                ns += gn == pn
+                both += (ga == pa) and (gn == pn)
+                if not ga and not pa:
+                    neg[name]["tp"] += 1
+                elif ga and not pa:
+                    neg[name]["fp"] += 1
+                elif not ga and pa:
+                    neg[name]["fn"] += 1
+            row[f"ab_{name}"] = ab / max(1, len(common))
+            row[f"nos_{name}"] = ns / max(1, len(common))
+            row[f"both_{name}"] = both / max(1, len(common))
+            tot[name]["ab"] += ab
+            tot[name]["nos"] += ns
+            tot[name]["both"] += both
+        tot["n"] += len(common)
+        rows.append(row)
+        print(f"[{s.key}] common {len(common)}: antibody acc {na} {row[f'ab_{na}']:.3f} {nb} {row[f'ab_{nb}']:.3f} | "
+              f"nostril acc {row[f'nos_{na}']:.3f} {row[f'nos_{nb}']:.3f} | both {row[f'both_{na}']:.3f} {row[f'both_{nb}']:.3f}", flush=True)
+    n = max(1, tot["n"])
+    L = [f"## Attribute accuracy on the {tot['n']} held-out nuclei matched by both {na} and {nb} (IoU >= 0.5, base run)", "",
+         f"| | {na} | {nb} |", "|---|---:|---:|"]
+    for k, label in (("ab", "Antibody +/− correct"), ("nos", "Nostril attribute correct"), ("both", "Both correct")):
+        L.append(f"| {label} | {100*tot[na][k]/n:.1f}% | {100*tot[nb][k]/n:.1f}% |")
+    for name in (na, nb):
+        t = neg[name]
+        p = t["tp"] / max(1, t["tp"] + t["fp"])
+        r = t["tp"] / max(1, t["tp"] + t["fn"])
+        f = 2 * p * r / max(1e-9, p + r)
+        L.append(f"| Antibody − on this set ({name}): TP / FP / FN, precision, recall, F1 | {t['tp']} / {t['fp']} / {t['fn']}, {100*p:.1f}%, {100*r:.1f}%, {100*f:.1f}% | |")
+    L += ["", f"| Image | matched by both | {na} antibody acc | {nb} antibody acc | {na} Nostril acc | {nb} Nostril acc |", "|---|---:|---:|---:|---:|---:|"]
+    for r in rows:
+        L.append(f"| {r['image']} | {r['n_common']} | {100*r[f'ab_{na}']:.1f}% | {100*r[f'ab_{nb}']:.1f}% | {100*r[f'nos_{na}']:.1f}% | {100*r[f'nos_{nb}']:.1f}% |")
+    L += ["", "## Counts summed over the nine held-out images (and summed absolute per-image count error)", "",
+          f"| Label | GT | {na} | {nb} | abs error {na} | abs error {nb} |", "|---|---:|---:|---:|---:|---:|"]
+    for lab in ALL_LABELS:
+        c = counts[lab]
+        L.append(f"| {lab} | {c['gt']} | {c[na]} | {c[nb]} | {c[f'abs_{na}']} | {c[f'abs_{nb}']} |")
+    L.append("")
+    text = "\n".join(L)
+    print(text)
+    if a.out:
+        with open(a.out, "a") as f:
+            f.write("\n" + text)
+        print("appended to", a.out)
+if __name__ == "__main__":
+    main()

scripts/compare_cv.py ADDED Viewed

	@@ -0,0 +1,110 @@

+#!/usr/bin/env python3
+"""Compare two cross-validation runs (cv_results.json files) on the measures that
+decide whether a new training recipe replaces the shipped weights.
+    python scripts/compare_cv.py reports/cv/cv_results.json runs/cv_v2/cv_results.json [--out reports/cv_v2/comparison.md]
+Nothing here looks at the images or the predictions; it only reads the two
+aggregates, so it cannot tune anything on the held-out data.
+"""
+from __future__ import annotations
+import argparse
+import json
+import os
+import sys
+sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+from cellquant.labels import ALL_LABELS  # noqa: E402
+def pct(x: float) -> str:
+    return f"{100 * x:.1f}%"
+def delta(a: float, b: float) -> str:
+    d = 100 * (b - a)
+    return f"{d:+.1f}"
+def headline_rows(agg: dict) -> dict[str, float]:
+    det, det8 = agg["detection_base"], agg["detection_tta8"]
+    rows = {
+        "Nucleus F1, IoU ≥ 0.5": det["cells"]["f1"],
+        "Nucleus precision, IoU ≥ 0.5": det["cells"]["precision"],
+        "Nucleus recall, IoU ≥ 0.5": det["cells"]["recall"],
+        "Nucleus F1, IoU ≥ 0.3": det["cells_iou30"]["f1"],
+        "Nucleus F1, centroid rule": det["cells_centroid"]["f1"],
+        "Nucleus F1, IoU ≥ 0.5, 8× TTA": det8["cells"]["f1"],
+        "Puncta F1": det["nostrils"]["f1"],
+        "Puncta F1, 8× TTA": det8["nostrils"]["f1"],
+    }
+    # attribute accuracies pooled over matched nuclei (weights = matched count per image)
+    rule = agg["rule_variants"]["rule_nos=head_cell=contains"]
+    rows["Attribute accuracy (both, matched nuclei)"] = rule["pooled_cell_attr_both"]
+    rows["Punctum type accuracy (matched puncta)"] = rule["pooled_nos_type_acc"]
+    for lab in ALL_LABELS:
+        rows[f"F1 [{lab}]"] = agg["micro_base"][lab]["f1"]
+    for lab in ("Cell, Antibody -",):
+        r = agg["micro_base"][lab]
+        rows[f"Precision [{lab}]"] = r["precision"]
+        rows[f"Recall [{lab}]"] = r["recall"]
+    return rows
+def main(argv=None):
+    ap = argparse.ArgumentParser()
+    ap.add_argument("old")
+    ap.add_argument("new")
+    ap.add_argument("--names", default="v1,v2")
+    ap.add_argument("--out", default="")
+    a = ap.parse_args(argv)
+    A = json.load(open(a.old))["aggregate"]
+    B = json.load(open(a.new))["aggregate"]
+    na, nb = a.names.split(",")
+    L = [f"# Cross-validation comparison: {na} vs {nb}", "",
+         f"`{a.old}` vs `{a.new}`; micro-averaged over the held-out images (base run, no TTA unless stated).", "",
+         f"| Measure | {na} | {nb} | Δ (points) |", "|---|---:|---:|---:|"]
+    ra, rb = headline_rows(A), headline_rows(B)
+    for k in ra:
+        L.append(f"| {k} | {pct(ra[k])} | {pct(rb[k])} | {delta(ra[k], rb[k])} |")
+    L += ["", "## Per held-out image", "",
+          f"| Image | Nuclei F1 {na} | Nuclei F1 {nb} | Δ | Centroid F1 {na} | Centroid F1 {nb} | Antibody acc {na} | Antibody acc {nb} | Puncta F1 {na} | Puncta F1 {nb} |",
+          "|---|---:|---:|---:|---:|---:|---:|---:|---:|---:|"]
+    for k in sorted(set(A["per_image"]) | set(B["per_image"])):
+        sa, sb = A["per_image"].get(k, {}).get("base"), B["per_image"].get(k, {}).get("base")
+        if not sa or not sb:
+            continue
+        L.append(f"| {k} | {pct(sa['cell_det_f1'])} | {pct(sb['cell_det_f1'])} | {delta(sa['cell_det_f1'], sb['cell_det_f1'])} | "
+                 f"{pct(sa['cell_det_f1_centroid'])} | {pct(sb['cell_det_f1_centroid'])} | {pct(sa['cell_attr_antibody'])} | {pct(sb['cell_attr_antibody'])} | "
+                 f"{pct(sa['nos_det_f1'])} | {pct(sb['nos_det_f1'])} |")
+    L += ["", "## Robustness (nuclei F1 over the 14 rotated / perturbed runs per image)", "",
+          f"| Image | {na} mean ± sd | {nb} mean ± sd |", "|---|---|---|"]
+    for k in sorted(set(A["per_image"]) & set(B["per_image"])):
+        ra_, rb_ = A["per_image"][k]["robustness"]["cell_det_f1"], B["per_image"][k]["robustness"]["cell_det_f1"]
+        L.append(f"| {k} | {100*ra_['mean']:.1f} ± {100*ra_['std']:.1f} | {100*rb_['mean']:.1f} ± {100*rb_['std']:.1f} |")
+    L += ["", "## Antibody-negative class", "",
+          f"| | {na} | {nb} |", "|---|---:|---:|"]
+    for lab in ("Cell, Antibody -", "Cell, Nostril, Antibody -"):
+        x, y = A["micro_base"][lab], B["micro_base"][lab]
+        L.append(f"| {lab}: GT / predicted / TP | {x['gt_count']} / {x['pred_count']} / {x['tp']} | {y['gt_count']} / {y['pred_count']} / {y['tp']} |")
+    sa, sb = A.get("antibody_threshold_sweep"), B.get("antibody_threshold_sweep")
+    if sa and sb:
+        ba = max(sa["sweep"], key=lambda r: r["neg_f1"])
+        bb = max(sb["sweep"], key=lambda r: r["neg_f1"])
+        at = [r for r in sa["sweep"] if abs(r["thr"] - 0.5) < 1e-6][0]
+        bt = [r for r in sb["sweep"] if abs(r["thr"] - 0.5) < 1e-6][0]
+        L.append(f"| Antibody − F1 among matched nuclei at the default threshold 0.5 | {pct(at['neg_f1'])} (P {pct(at['neg_precision'])}, R {pct(at['neg_recall'])}) | {pct(bt['neg_f1'])} (P {pct(bt['neg_precision'])}, R {pct(bt['neg_recall'])}) |")
+        L.append(f"| Best threshold in the sweep (not used) | {pct(ba['neg_f1'])} at {ba['thr']:.2f} | {pct(bb['neg_f1'])} at {bb['thr']:.2f} |")
+    L.append("")
+    text = "\n".join(L)
+    print(text)
+    if a.out:
+        os.makedirs(os.path.dirname(a.out), exist_ok=True)
+        with open(a.out, "w") as f:
+            f.write(text)
+        print("wrote", a.out)
+if __name__ == "__main__":
+    main()

tests/test_data.py CHANGED Viewed

@@ -19,8 +19,9 @@ def test_make_targets_layout():
     T = make_targets(_raster())
     assert set(np.unique(T["cell3"])) == {0, 1, 2}
     assert ((T["cell3"] > 0) == (T["cellmask"] > 0)).all()
-    # border ring is 2 px wide: a 20x20 square has 20*20 - 16*16 = 144 border px
-    assert int((T["cell3"][10:30, 10:30] == 2).sum()) == 144
     assert T["antibody"][15, 15] == 1 and T["antibody"][40, 40] == 0
     assert T["cellnos"][15, 15] == 1 and T["cellnos"][40, 40] == 0
     assert T["nos3"][21, 21] == 1 and T["nos3"][5, 50] == 2 and T["nos3"][0, 0] == 0

     T = make_targets(_raster())
     assert set(np.unique(T["cell3"])) == {0, 1, 2}
     assert ((T["cell3"] > 0) == (T["cellmask"] > 0)).all()
+    # border ring is 3 px wide: a 20x20 square has 20*20 - 14*14 = 204 border px
+    assert int((T["cell3"][10:30, 10:30] == 2).sum()) == 204
+    assert T["dist"].shape == T["cell3"].shape and T["dist"].dtype == np.float32
     assert T["antibody"][15, 15] == 1 and T["antibody"][40, 40] == 0
     assert T["cellnos"][15, 15] == 1 and T["cellnos"][40, 40] == 0
     assert T["nos3"][21, 21] == 1 and T["nos3"][5, 50] == 2 and T["nos3"][0, 0] == 0

tests/test_v2_recipe.py ADDED Viewed

	@@ -0,0 +1,152 @@

+"""Tests for the v2 recipe: distance channel, 3-px border, balanced antibody loss,
+oversampling of antibody-negative nuclei, and backward compatibility with 8-channel
+checkpoints."""
+import numpy as np
+import torch
+from cellquant.annotations import RasterAnnotation
+from cellquant.data import AugConfig, PatchSampler, Prefetcher, distance_target, make_targets, negative_nucleus_points
+from cellquant.infer import Predictor
+from cellquant.losses import LAMBDA_DIST, masked_bce, multitask_loss
+from cellquant.model import DIST_CH, N_OUT, N_OUT_LEGACY, CellUNet, load_checkpoint, save_checkpoint
+from cellquant.postprocess import PostConfig, analyze, logits_to_probs, segment_cells
+def _disc(h, w, cy, cx, r):
+    yy, xx = np.mgrid[:h, :w]
+    return (yy - cy) ** 2 + (xx - cx) ** 2 <= r * r
+def test_distance_target_peaks_at_one_and_dips_between_touching_nuclei():
+    inst = np.zeros((40, 60), np.int32)
+    inst[10:30, 10:30] = 1
+    inst[10:30, 30:50] = 2  # shares a full edge with instance 1
+    d = distance_target(inst)
+    assert d.dtype == np.float32 and d.min() == 0 and (d[inst == 0] == 0).all()
+    assert np.isclose(d[inst == 1].max(), 1.0) and np.isclose(d[inst == 2].max(), 1.0)
+    # the contact line is as low as the outer outline: 1 px / 10 px
+    assert np.isclose(d[20, 29], 0.1) and np.isclose(d[20, 30], 0.1) and np.isclose(d[20, 10], 0.1)
+    assert d[20, 19] > 0.9 and d[20, 40] > 0.9
+    # the image edge counts as outside
+    edge = np.zeros((20, 20), np.int32)
+    edge[0:10, 0:10] = 1
+    de = distance_target(edge)
+    assert np.isclose(de[0, 0], 0.2) and np.isclose(de[4, 4], 1.0) and np.isclose(de[9, 9], 0.2)
+def test_distance_watershed_splits_what_the_interior_map_merges():
+    h = w = 96
+    d1, d2 = _disc(h, w, 48, 30, 14), _disc(h, w, 48, 56, 14)
+    fg = d1 | d2
+    cell = np.zeros((3, h, w), np.float32)
+    cell[1][fg] = 0.9           # one solid interior blob, no border between the discs
+    cell[0] = 1 - cell[1]
+    inst_plain = segment_cells(cell, PostConfig())
+    assert inst_plain.max() == 1  # merged without the distance map
+    gt = np.zeros((h, w), np.int32)
+    gt[d1] = 1
+    gt[d2 & ~d1] = 2
+    dist = distance_target(gt)
+    inst_dist = segment_cells(cell, PostConfig(), dist)
+    assert inst_dist.max() == 2
+    areas = np.bincount(inst_dist.ravel())[1:]
+    assert all(450 <= a <= 750 for a in areas)
+    # analyze() picks the distance map up from the probability dict
+    probs = {"cell": cell, "antibody": np.full((h, w), 0.9, np.float32), "cellnos": np.zeros((h, w), np.float32),
+             "nos": np.stack([np.ones((h, w)), np.zeros((h, w)), np.zeros((h, w))]).astype(np.float32), "dist": dist}
+    assert len(analyze(probs, PostConfig()).cells) == 2
+    del probs["dist"]
+    assert len(analyze(probs, PostConfig()).cells) == 1
+def test_logits_to_probs_with_and_without_distance_channel():
+    rng = np.random.default_rng(0)
+    p8 = logits_to_probs(rng.normal(size=(8, 16, 16)).astype(np.float32))
+    assert "dist" not in p8
+    p9 = logits_to_probs(rng.normal(size=(9, 16, 16)).astype(np.float32) * 3)
+    assert p9["dist"].shape == (16, 16) and p9["dist"].min() >= 0 and p9["dist"].max() <= 1
+    assert np.allclose(p9["cell"].sum(0), 1, atol=1e-5)
+def test_predictor_old_and_new_checkpoints(tmp_path):
+    img = (np.random.default_rng(0).random((70, 90, 3)) * 80).astype(np.uint8)
+    old = CellUNet("resnet34", pretrained=False, n_out=N_OUT_LEGACY).eval()
+    save_checkpoint(old, str(tmp_path / "old.pt"))
+    m_old, meta = load_checkpoint(str(tmp_path / "old.pt"))
+    assert meta["n_out"] == N_OUT_LEGACY and m_old.head.out_channels == N_OUT_LEGACY
+    p_old = Predictor(str(tmp_path / "old.pt"), device="cpu")
+    pr = p_old.probs(img)
+    assert not p_old.has_dist and "dist" not in pr and pr["cell"].shape == (3, 70, 90)
+    new = CellUNet("resnet34", pretrained=False).eval()
+    assert new.head.out_channels == N_OUT == DIST_CH + 1
+    p_new = Predictor(new, device="cpu")
+    pr = p_new.probs(img)
+    assert p_new.has_dist and pr["dist"].shape == (70, 90) and pr["dist"].min() >= 0 and pr["dist"].max() <= 1
+    # tiled path and TTA keep the channel
+    tiled = Predictor(new, device="cpu", tile=64, overlap=16, max_whole=32).probs(img)
+    assert tiled["dist"].shape == (70, 90)
+    assert Predictor(new, device="cpu", tta=4).probs(img)["dist"].shape == (70, 90)
+    # a whole prediction runs end to end on the new layout
+    res, probs = p_new.predict(img, PostConfig())
+    assert "dist" in probs and isinstance(res.cells, list)
+def test_loss_has_distance_term_and_negative_weight_balances_antibody():
+    rng = np.random.default_rng(0)
+    B, H, W = 2, 32, 32
+    inst = np.zeros((H, W), np.int32)
+    inst[4:14, 4:14] = 1
+    inst[18:28, 18:28] = 2
+    R = RasterAnnotation(inst, ["", "Cell, Antibody +", "Cell, Antibody -"], np.zeros((H, W), np.int32), [""])
+    T = make_targets(R)
+    t = {k: torch.from_numpy(np.stack([v, v])) for k, v in T.items()}
+    out9 = torch.from_numpy(rng.normal(size=(B, 9, H, W)).astype(np.float32))
+    total9, parts9 = multitask_loss(out9, t)
+    assert "mse_dist" in parts9 and parts9["mse_dist"] > 0
+    assert np.isclose(parts9["total"], sum(v for k, v in parts9.items() if k not in ("total", "mse_dist")) + LAMBDA_DIST * parts9["mse_dist"], atol=1e-4)
+    out8 = out9[:, :8]
+    _, parts8 = multitask_loss(out8, t)
+    assert "mse_dist" not in parts8
+    # a perfect distance prediction removes the term
+    perfect = out9.clone()
+    perfect[:, DIST_CH] = t["dist"]
+    assert multitask_loss(perfect, t)[1]["mse_dist"] < 1e-6
+    # the antibody head is confidently positive everywhere: up-weighting negatives raises the loss
+    logit = torch.full((B, H, W), 3.0)
+    mask = t["cellmask"]
+    l1 = masked_bce(logit, t["antibody"], mask, neg_weight=1.0)
+    l5 = masked_bce(logit, t["antibody"], mask, neg_weight=5.0)
+    assert l5 > l1
+    # ... and leaves an all-positive batch unchanged
+    assert torch.isclose(masked_bce(logit, torch.ones_like(logit), mask, neg_weight=5.0),
+                         masked_bce(logit, torch.ones_like(logit), mask, neg_weight=1.0))
+def test_focus_sampling_puts_a_negative_nucleus_in_every_crop():
+    H = W = 128
+    inst = np.zeros((H, W), np.int32)
+    inst[100:110, 10:20] = 1    # antibody-negative nucleus in one corner
+    inst[20:30, 90:100] = 2     # antibody-positive one in the opposite corner
+    R = RasterAnnotation(inst, ["", "Cell, Antibody -", "Cell, Antibody +"], np.zeros((H, W), np.int32), [""])
+    pts = negative_nucleus_points(R)
+    assert pts.shape == (1, 2) and np.allclose(pts[0], (104.5, 14.5))
+    assert negative_nucleus_points(RasterAnnotation(inst, ["", "Cell, Antibody +", "Cell, Antibody +"], np.zeros((H, W), np.int32), [""])).shape == (0, 2)
+    img = (np.random.default_rng(0).random((H, W, 3)) * 255).astype(np.uint8)
+    T = make_targets(R)
+    cfg = AugConfig(crop=32, p_focus=1.0)
+    _, t = PatchSampler([img], [T], cfg, seed=1, focus=[pts]).batch(16)
+    neg = (t["antibody"] < 0.5) & (t["cellmask"] > 0)
+    assert all(neg[i].any() for i in range(16))
+    # without focus points (or p_focus = 0) crops are placed at random, so the 10x10 nucleus is often missed
+    _, t0 = PatchSampler([img], [T], AugConfig(crop=32, p_focus=0.0), seed=1, focus=[pts]).batch(16)
+    neg0 = (t0["antibody"] < 0.5) & (t0["cellmask"] > 0)
+    assert sum(neg0[i].any() for i in range(16)) < 16
+    # the prefetcher passes the points through and stays deterministic
+    def stream(seed):
+        pf = Prefetcher([img], [T], cfg, batch=2, seed=seed, workers=2, depth=4, focus=[pts])
+        out = [pf.next()[1]["antibody"].numpy().copy() for _ in range(4)]
+        pf.close()
+        return out
+    a, b = stream(2), stream(2)
+    assert all(np.array_equal(x, y) for x, y in zip(a, b))
+    assert all(((x < 0.5) & (x >= 0)).any() for x in a)  # cannot see cellmask here, but focus crops always contain the nucleus

train.py CHANGED Viewed

@@ -22,7 +22,7 @@ import torch
 sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
 from cellquant.annotations import discover, load_image, rasterize  # noqa: E402
-from cellquant.data import AugConfig, Prefetcher, make_targets  # noqa: E402
 from cellquant.infer import Predictor, pick_device  # noqa: E402
 from cellquant.losses import multitask_loss  # noqa: E402
 from cellquant.metrics import evaluate_cells, evaluate_nostrils  # noqa: E402
@@ -48,9 +48,24 @@ def parse_args(argv=None):
     ap.add_argument("--log-every", type=int, default=50)
     ap.add_argument("--val-every", type=int, default=500)
     ap.add_argument("--workers", type=int, default=3, help="augmentation threads")
     return ap.parse_args(argv)
 def lr_at(it: int, total: int, base: float, warmup: int) -> float:
     if it < warmup:
         return base * (it + 1) / warmup
@@ -85,13 +100,18 @@ def main(argv=None):
     if a.holdout and not held:
         raise SystemExit(f"holdout {a.holdout!r} not among {[s.key for s in samples]}")
     print(f"device={device} train={[s.key for s in train]} holdout={[s.key for s in held]}")
-    images, targets = [], []
     for s in train:
         img = load_image(s.image_path)
         R = rasterize(s.annotation, img.shape[0], img.shape[1])
         images.append(img)
         targets.append(make_targets(R))
-    sampler = Prefetcher(images, targets, AugConfig(crop=a.crop), batch=a.batch, seed=a.seed, workers=a.workers)
     held_img = held_R = None
     if held:
         held_img = load_image(held[0].image_path)
@@ -109,7 +129,7 @@ def main(argv=None):
         model.train()
         x, t = sampler.next()
         out = model(x.to(device))
-        loss, parts = multitask_loss(out, t)
         opt.zero_grad(set_to_none=True)
         loss.backward()
         torch.nn.utils.clip_grad_norm_(model.parameters(), 5.0)
@@ -125,7 +145,7 @@ def main(argv=None):
             history["train"].append(avg)
             print(f"it {it+1:5d} loss {avg['total']:.3f} cell {avg['ce_cell']:.3f}/{avg['dice_cell']:.3f} "
                   f"nos {avg['ce_nos']:.3f}/{avg['dice_nos']:.3f} ab {avg['bce_ab']:.3f} cn {avg['bce_cn']:.3f} "
-                  f"lr {lr:.2e} {avg['elapsed_s']:.0f}s", flush=True)
         if held and ((it + 1) % a.val_every == 0 or it + 1 == a.iters):
             model.eval()
             v = quick_val(model, held_img, held_R, device)
@@ -137,7 +157,7 @@ def main(argv=None):
     model.eval().cpu()
     os.makedirs(os.path.dirname(os.path.abspath(a.out)), exist_ok=True)
     save_checkpoint(model, a.out, {
-        "config": vars(a),
         "train_keys": [s.key for s in train],
         "holdout": a.holdout,
         "history": history,

 sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
 from cellquant.annotations import discover, load_image, rasterize  # noqa: E402
+from cellquant.data import AugConfig, Prefetcher, make_targets, negative_nucleus_points  # noqa: E402
 from cellquant.infer import Predictor, pick_device  # noqa: E402
 from cellquant.losses import multitask_loss  # noqa: E402
 from cellquant.metrics import evaluate_cells, evaluate_nostrils  # noqa: E402
     ap.add_argument("--log-every", type=int, default=50)
     ap.add_argument("--val-every", type=int, default=500)
     ap.add_argument("--workers", type=int, default=3, help="augmentation threads")
+    ap.add_argument("--p-focus", type=float, default=AugConfig.p_focus,
+                    help="share of crops forced to contain an antibody-negative nucleus (0 = off)")
+    ap.add_argument("--ab-neg-weight", default="auto",
+                    help="loss weight of antibody-negative nucleus pixels: 'auto' = sqrt(pos px / neg px) clipped to [1, 5], or a number")
     return ap.parse_args(argv)
+def antibody_negative_weight(targets: list[dict], spec: str = "auto") -> float:
+    """Class-balancing weight for the antibody head (see losses.masked_bce)."""
+    if spec != "auto":
+        return float(spec)
+    pos = sum(float(((t["antibody"] > 0.5) & (t["cellmask"] > 0)).sum()) for t in targets)
+    neg = sum(float(((t["antibody"] < 0.5) & (t["cellmask"] > 0)).sum()) for t in targets)
+    if neg == 0:
+        return 1.0
+    return float(np.clip(np.sqrt(pos / neg), 1.0, 5.0))
 def lr_at(it: int, total: int, base: float, warmup: int) -> float:
     if it < warmup:
         return base * (it + 1) / warmup
     if a.holdout and not held:
         raise SystemExit(f"holdout {a.holdout!r} not among {[s.key for s in samples]}")
     print(f"device={device} train={[s.key for s in train]} holdout={[s.key for s in held]}")
+    images, targets, focus = [], [], []
     for s in train:
         img = load_image(s.image_path)
         R = rasterize(s.annotation, img.shape[0], img.shape[1])
         images.append(img)
         targets.append(make_targets(R))
+        focus.append(negative_nucleus_points(R))
+    ab_neg_weight = antibody_negative_weight(targets, a.ab_neg_weight)
+    print(f"antibody-negative nuclei in training set: {sum(len(f) for f in focus)} "
+          f"(loss weight {ab_neg_weight:.2f}, focus share {a.p_focus:.2f})")
+    sampler = Prefetcher(images, targets, AugConfig(crop=a.crop, p_focus=a.p_focus), batch=a.batch, seed=a.seed,
+                         workers=a.workers, focus=focus if a.p_focus > 0 else None)
     held_img = held_R = None
     if held:
         held_img = load_image(held[0].image_path)
         model.train()
         x, t = sampler.next()
         out = model(x.to(device))
+        loss, parts = multitask_loss(out, t, ab_neg_weight=ab_neg_weight)
         opt.zero_grad(set_to_none=True)
         loss.backward()
         torch.nn.utils.clip_grad_norm_(model.parameters(), 5.0)
             history["train"].append(avg)
             print(f"it {it+1:5d} loss {avg['total']:.3f} cell {avg['ce_cell']:.3f}/{avg['dice_cell']:.3f} "
                   f"nos {avg['ce_nos']:.3f}/{avg['dice_nos']:.3f} ab {avg['bce_ab']:.3f} cn {avg['bce_cn']:.3f} "
+                  f"dist {avg.get('mse_dist', 0.0):.4f} lr {lr:.2e} {avg['elapsed_s']:.0f}s", flush=True)
         if held and ((it + 1) % a.val_every == 0 or it + 1 == a.iters):
             model.eval()
             v = quick_val(model, held_img, held_R, device)
     model.eval().cpu()
     os.makedirs(os.path.dirname(os.path.abspath(a.out)), exist_ok=True)
     save_checkpoint(model, a.out, {
+        "config": {**vars(a), "ab_neg_weight_used": ab_neg_weight},
         "train_keys": [s.key for s in train],
         "holdout": a.holdout,
         "history": history,

weights/cellquant.pt CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b6d252d9f4c03aedcf717649cc766fb47a503acc7cca8135b8d43bc684838959
-size 98290507

 version https://git-lfs.github.com/spec/v1
+oid sha256:0a9ea444de463739cab0de9b17061fb9931fa338d8975c984c23cc826fc35042
+size 98292471

weights/cellquant_history.json CHANGED Viewed

@@ -1,724 +1,784 @@
 {
  "train": [
   {
-   "ce_cell": 0.9070329773426056,
-   "dice_cell": 0.8730328452587127,
-   "ce_nos": 1.0830040109157562,
-   "dice_nos": 0.9898272252082825,
-   "bce_ab": 0.524188295006752,
-   "bce_cn": 0.6183760815858841,
-   "total": 4.9954613590240475,
    "iter": 50,
    "lr": 0.00015,
-   "elapsed_s": 59.566773891448975
   },
   {
-   "ce_cell": 0.7300815117359162,
-   "dice_cell": 0.8212458372116089,
-   "ce_nos": 0.7789127671718598,
-   "dice_nos": 0.9821323096752167,
-   "bce_ab": 0.2914739960432053,
-   "bce_cn": 0.3936499571800232,
-   "total": 3.9974963092803955,
    "iter": 100,
    "lr": 0.0003,
-   "elapsed_s": 109.73091292381287
   },
   {
-   "ce_cell": 0.5734736859798432,
-   "dice_cell": 0.7798113775253296,
-   "ce_nos": 0.574396123290062,
-   "dice_nos": 0.9748807692527771,
-   "bce_ab": 0.23322827845811844,
-   "bce_cn": 0.33575250566005704,
-   "total": 3.4715427684783937,
    "iter": 150,
    "lr": 0.00029979294712266676,
-   "elapsed_s": 160.73545289039612
   },
   {
-   "ce_cell": 0.44446682393550874,
-   "dice_cell": 0.7253605389595031,
-   "ce_nos": 0.4329249274730682,
-   "dice_nos": 0.9701189231872559,
-   "bce_ab": 0.20956841707229615,
-   "bce_cn": 0.32494749456644056,
-   "total": 3.107387113571167,
    "iter": 200,
    "lr": 0.0002991554114956718,
-   "elapsed_s": 211.49982213974
   },
   {
-   "ce_cell": 0.35687846958637237,
-   "dice_cell": 0.6721045792102813,
-   "ce_nos": 0.32840390503406525,
-   "dice_nos": 0.9639902281761169,
-   "bce_ab": 0.17408027544617652,
-   "bce_cn": 0.3147788044810295,
-   "total": 2.810236248970032,
    "iter": 250,
    "lr": 0.00029808917686451123,
-   "elapsed_s": 255.0301797389984
   },
   {
-   "ce_cell": 0.30268543839454654,
-   "dice_cell": 0.6246984755992889,
-   "ce_nos": 0.24472756624221803,
-   "dice_nos": 0.9488238263130188,
-   "bce_ab": 0.1679477596282959,
-   "bce_cn": 0.264477858543396,
-   "total": 2.5533609533309938,
    "iter": 300,
    "lr": 0.00029659737067906796,
-   "elapsed_s": 305.7051999568939
   },
   {
-   "ce_cell": 0.2692413151264191,
-   "dice_cell": 0.5903848171234131,
-   "ce_nos": 0.18806805282831193,
-   "dice_nos": 0.9169448506832123,
-   "bce_ab": 0.1362369504570961,
-   "bce_cn": 0.2647552129626274,
-   "total": 2.365631227493286,
    "iter": 350,
    "lr": 0.00029468436866391876,
-   "elapsed_s": 356.85323691368103
   },
   {
-   "ce_cell": 0.24572547733783723,
-   "dice_cell": 0.5541446590423584,
-   "ce_nos": 0.14375351637601852,
-   "dice_nos": 0.8989665198326111,
-   "bce_ab": 0.1442735657095909,
-   "bce_cn": 0.25428884506225585,
-   "total": 2.241152620315552,
    "iter": 400,
    "lr": 0.0002923557819835801,
-   "elapsed_s": 408.5768549442291
   },
   {
-   "ce_cell": 0.23625801086425782,
-   "dice_cell": 0.5364066225290298,
-   "ce_nos": 0.11115477636456489,
-   "dice_nos": 0.87745032787323,
-   "bce_ab": 0.1183611673116684,
-   "bce_cn": 0.22456467717885972,
-   "total": 2.1041955590248107,
    "iter": 450,
    "lr": 0.00028961844078399557,
-   "elapsed_s": 459.0741550922394
   },
   {
-   "ce_cell": 0.21822448045015336,
-   "dice_cell": 0.5127934575080871,
-   "ce_nos": 0.0872543377429247,
-   "dice_nos": 0.8334435498714448,
-   "bce_ab": 0.1045101822167635,
-   "bce_cn": 0.2290149673819542,
-   "total": 1.9852409744262696,
    "iter": 500,
    "lr": 0.0002864803741585413,
-   "elapsed_s": 509.7258110046387
   },
   {
-   "ce_cell": 0.21543064922094346,
-   "dice_cell": 0.49275926291942596,
-   "ce_nos": 0.07010223433375358,
-   "dice_nos": 0.795907906293869,
-   "bce_ab": 0.10423918575048446,
-   "bce_cn": 0.21024613380432128,
-   "total": 1.8886853814125062,
    "iter": 550,
    "lr": 0.0002829507865973116,
-   "elapsed_s": 560.3340859413147
   },
   {
-   "ce_cell": 0.21174423784017563,
-   "dice_cell": 0.4891924291849136,
-   "ce_nos": 0.06100128136575222,
-   "dice_nos": 0.7548426115512847,
-   "bce_ab": 0.09695948652923107,
-   "bce_cn": 0.20541400745511054,
-   "total": 1.8191540384292602,
    "iter": 600,
    "lr": 0.0002790400309887635,
-   "elapsed_s": 610.7946240901947
   },
   {
-   "ce_cell": 0.20390785247087478,
-   "dice_cell": 0.4746441453695297,
-   "ce_nos": 0.04720834020525217,
-   "dice_nos": 0.709533805847168,
-   "bce_ab": 0.09941868580877782,
-   "bce_cn": 0.19302350237965585,
-   "total": 1.727736325263977,
    "iter": 650,
    "lr": 0.0002747595782529114,
-   "elapsed_s": 660.7677178382874
   },
   {
-   "ce_cell": 0.2039007693529129,
-   "dice_cell": 0.46679477393627167,
-   "ce_nos": 0.048075792603194716,
-   "dice_nos": 0.6227267402410507,
-   "bce_ab": 0.08658930584788323,
-   "bce_cn": 0.18947327509522438,
-   "total": 1.617560646533966,
    "iter": 700,
    "lr": 0.0002701219836951418,
-   "elapsed_s": 711.0392649173737
   },
   {
-   "ce_cell": 0.200768101811409,
-   "dice_cell": 0.45868332326412203,
-   "ce_nos": 0.03979761004447937,
-   "dice_nos": 0.5989265680313111,
-   "bce_ab": 0.07302961058914662,
-   "bce_cn": 0.1712094095349312,
-   "total": 1.5424146223068238,
    "iter": 750,
    "lr": 0.0002651408501793399,
-   "elapsed_s": 760.5930259227753
   },
   {
-   "ce_cell": 0.19177152156829835,
-   "dice_cell": 0.4472357839345932,
-   "ce_nos": 0.03094205627217889,
-   "dice_nos": 0.5603820198774337,
-   "bce_ab": 0.08034474328160286,
-   "bce_cn": 0.14997318997979164,
-   "total": 1.4606493067741395,
    "iter": 800,
    "lr": 0.0002598307882283455,
-   "elapsed_s": 810.9303019046783
   },
   {
-   "ce_cell": 0.19938127607107162,
-   "dice_cell": 0.44567417323589326,
-   "ce_nos": 0.04125866884365678,
-   "dice_nos": 0.4803364199399948,
-   "bce_ab": 0.07717311266809702,
-   "bce_cn": 0.1591106753051281,
-   "total": 1.4029343223571777,
    "iter": 850,
    "lr": 0.00025420737316877263,
-   "elapsed_s": 857.9806337356567
   },
   {
-   "ce_cell": 0.1933874309062958,
-   "dice_cell": 0.4387037259340286,
-   "ce_nos": 0.036275244057178496,
-   "dice_nos": 0.48662609457969663,
-   "bce_ab": 0.06898861289024354,
-   "bce_cn": 0.15267845779657363,
-   "total": 1.3766595673561097,
    "iter": 900,
    "lr": 0.00024828709944589296,
-   "elapsed_s": 908.7008759975433
   },
   {
-   "ce_cell": 0.19506525486707688,
-   "dice_cell": 0.4412110126018524,
-   "ce_nos": 0.035951650235801935,
-   "dice_nos": 0.43454607129096984,
-   "bce_ab": 0.06932192835956812,
-   "bce_cn": 0.14733665704727172,
-   "total": 1.32343257188797,
    "iter": 950,
    "lr": 0.00024208733224258537,
-   "elapsed_s": 959.4800848960876
   },
   {
-   "ce_cell": 0.1945480227470398,
-   "dice_cell": 0.43224972307682036,
-   "ce_nos": 0.0335944464802742,
-   "dice_nos": 0.40378140449523925,
-   "bce_ab": 0.06522910814732313,
-   "bce_cn": 0.13741043150424959,
-   "total": 1.2668131160736085,
    "iter": 1000,
    "lr": 0.0002356262565442632,
-   "elapsed_s": 1009.7932658195496
   },
   {
-   "ce_cell": 0.18734178960323333,
-   "dice_cell": 0.4267615270614624,
-   "ce_nos": 0.036363077107816937,
-   "dice_nos": 0.4053969383239746,
-   "bce_ab": 0.0643582896143198,
-   "bce_cn": 0.12727220386266708,
-   "total": 1.2474938321113587,
    "iter": 1050,
    "lr": 0.0002289228237991784,
-   "elapsed_s": 1059.9658198356628
   },
   {
-   "ce_cell": 0.18910773277282714,
-   "dice_cell": 0.42362256467342374,
-   "ce_nos": 0.0346462229359895,
-   "dice_nos": 0.3729756760597229,
-   "bce_ab": 0.06174593850970268,
-   "bce_cn": 0.13085844799876212,
-   "total": 1.2129565787315368,
    "iter": 1100,
    "lr": 0.0002219966963305587,
-   "elapsed_s": 1110.0916857719421
   },
   {
-   "ce_cell": 0.1822684535384178,
-   "dice_cell": 0.4143069595098495,
-   "ce_nos": 0.03322938298806548,
-   "dice_nos": 0.3389162081480026,
-   "bce_ab": 0.05988596722483635,
-   "bce_cn": 0.11884681917726994,
-   "total": 1.1474537920951844,
    "iter": 1150,
    "lr": 0.00021486818966362585,
-   "elapsed_s": 1160.7585599422455
   },
   {
-   "ce_cell": 0.18991790741682052,
-   "dice_cell": 0.41873910784721374,
-   "ce_nos": 0.036337544079869985,
-   "dice_nos": 0.34970403611660006,
-   "bce_ab": 0.05970212183892727,
-   "bce_cn": 0.1334019273519516,
-   "total": 1.1878026413917542,
    "iter": 1200,
    "lr": 0.00020755821293665946,
-   "elapsed_s": 1211.4072179794312
   },
   {
-   "ce_cell": 0.17697676301002502,
-   "dice_cell": 0.40928772032260896,
-   "ce_nos": 0.03304771558381617,
-   "dice_nos": 0.3399574244022369,
-   "bce_ab": 0.04348419822752476,
-   "bce_cn": 0.1121597220748663,
-   "total": 1.1149135494232179,
    "iter": 1250,
    "lr": 0.00020008820757089248,
-   "elapsed_s": 1262.1290378570557
   },
   {
-   "ce_cell": 0.17598221957683563,
-   "dice_cell": 0.40642625391483306,
-   "ce_nos": 0.033407045025378466,
-   "dice_nos": 0.310367751121521,
-   "bce_ab": 0.0478444230183959,
-   "bce_cn": 0.09985558047890664,
-   "total": 1.073883285522461,
    "iter": 1300,
    "lr": 0.0001924800843791277,
-   "elapsed_s": 1311.735403060913
   },
   {
-   "ce_cell": 0.18522560983896255,
-   "dice_cell": 0.403508597612381,
-   "ce_nos": 0.036090935450047254,
-   "dice_nos": 0.3127438998222351,
-   "bce_ab": 0.04713532615453005,
-   "bce_cn": 0.11126945935189723,
-   "total": 1.0959738206863403,
    "iter": 1350,
    "lr": 0.00018475615929754922,
-   "elapsed_s": 1362.2759177684784
   },
   {
-   "ce_cell": 0.18232185065746306,
-   "dice_cell": 0.40336430847644805,
-   "ce_nos": 0.0384017390338704,
-   "dice_nos": 0.30273705661296846,
-   "bce_ab": 0.04570051861926913,
-   "bce_cn": 0.11299202486872673,
-   "total": 1.0855175077915191,
    "iter": 1400,
    "lr": 0.0001769390879292361,
-   "elapsed_s": 1412.6486489772797
   },
   {
-   "ce_cell": 0.18443019971251487,
-   "dice_cell": 0.41006422102451323,
-   "ce_nos": 0.038777465261518956,
-   "dice_nos": 0.293556307554245,
-   "bce_ab": 0.040375117510557175,
-   "bce_cn": 0.0971392347663641,
-   "total": 1.0643425357341767,
    "iter": 1450,
    "lr": 0.00016905179909137546,
-   "elapsed_s": 1462.6795949935913
   },
   {
-   "ce_cell": 0.17869871109724045,
-   "dice_cell": 0.40553332567214967,
-   "ce_nos": 0.03795233623124659,
-   "dice_nos": 0.2916260594129562,
-   "bce_ab": 0.04099308693781495,
-   "bce_cn": 0.0865385101735592,
-   "total": 1.0413420403003693,
    "iter": 1500,
    "lr": 0.00016111742756109098,
-   "elapsed_s": 1512.4294548034668
   },
   {
-   "ce_cell": 0.1764592757821083,
-   "dice_cell": 0.3958728986978531,
-   "ce_nos": 0.035169387739151715,
-   "dice_nos": 0.27471464276313784,
-   "bce_ab": 0.035233155209571126,
-   "bce_cn": 0.08008314736187458,
-   "total": 0.9975325107574463,
    "iter": 1550,
    "lr": 0.00015315924621715527,
-   "elapsed_s": 1562.1074118614197
   },
   {
-   "ce_cell": 0.17475848585367204,
-   "dice_cell": 0.3929943698644638,
-   "ce_nos": 0.03353961421176791,
-   "dice_nos": 0.3009433788061142,
-   "bce_ab": 0.03854949928820133,
-   "bce_cn": 0.08339719414710998,
-   "total": 1.0241825413703918,
    "iter": 1600,
    "lr": 0.00014520059777662684,
-   "elapsed_s": 1612.9562838077545
   },
   {
-   "ce_cell": 0.18072757810354234,
-   "dice_cell": 0.39913213849067686,
-   "ce_nos": 0.039328129449859264,
-   "dice_nos": 0.2857843828201294,
-   "bce_ab": 0.03886710917577148,
-   "bce_cn": 0.09502572849392892,
-   "total": 1.0388650584220886,
    "iter": 1650,
    "lr": 0.00013726482632663912,
-   "elapsed_s": 1663.1288919448853
   },
   {
-   "ce_cell": 0.1759936460852623,
-   "dice_cell": 0.39650719165802,
-   "ce_nos": 0.032182310023345054,
-   "dice_nos": 0.2933516526222229,
-   "bce_ab": 0.03653882872313261,
-   "bce_cn": 0.08788134880363942,
-   "total": 1.0224549758434296,
    "iter": 1700,
    "lr": 0.00012937520885217038,
-   "elapsed_s": 1712.6260061264038
   },
   {
-   "ce_cell": 0.17498762905597687,
-   "dice_cell": 0.38849363803863524,
-   "ce_nos": 0.03151271438226104,
-   "dice_nos": 0.27745749831199645,
-   "bce_ab": 0.038630726784467696,
-   "bce_cn": 0.08467736341059208,
-   "total": 0.995759572982788,
    "iter": 1750,
    "lr": 0.0001215548869606366,
-   "elapsed_s": 1763.162414073944
   },
   {
-   "ce_cell": 0.16943098440766335,
-   "dice_cell": 0.38974578082561495,
-   "ce_nos": 0.030594841404818,
-   "dice_nos": 0.2775198197364807,
-   "bce_ab": 0.03063900684937835,
-   "bce_cn": 0.07809612043201923,
-   "total": 0.9760265576839448,
    "iter": 1800,
    "lr": 0.00011382679900356957,
-   "elapsed_s": 1812.9093608856201
   },
   {
-   "ce_cell": 0.17392412394285203,
-   "dice_cell": 0.39087122797966,
-   "ce_nos": 0.03287358325906098,
-   "dice_nos": 0.27309031009674073,
-   "bce_ab": 0.03879237989895046,
-   "bce_cn": 0.07375255063176155,
-   "total": 0.9833041846752166,
    "iter": 1850,
    "lr": 0.0001062136127944796,
-   "elapsed_s": 1862.8471250534058
   },
   {
-   "ce_cell": 0.1721265098452568,
-   "dice_cell": 0.3918482345342636,
-   "ce_nos": 0.03052111998666078,
-   "dice_nos": 0.26571854174137116,
-   "bce_ab": 0.028863372541964055,
-   "bce_cn": 0.07009109571576118,
-   "total": 0.9591688740253449,
    "iter": 1900,
    "lr": 9.873765912025276e-05,
-   "elapsed_s": 1913.0803048610687
   },
   {
-   "ce_cell": 0.17321533888578414,
-   "dice_cell": 0.3879319250583649,
-   "ce_nos": 0.03197065497748554,
-   "dice_nos": 0.2754987716674805,
-   "bce_ab": 0.03000295827165246,
-   "bce_cn": 0.0681322829797864,
-   "total": 0.966751925945282,
    "iter": 1950,
    "lr": 9.142086624110557e-05,
-   "elapsed_s": 1962.4520688056946
   },
   {
-   "ce_cell": 0.16933307886123658,
-   "dice_cell": 0.3823584771156311,
-   "ce_nos": 0.03379744639620185,
-   "dice_nos": 0.2684609520435333,
-   "bce_ab": 0.03405721362680197,
-   "bce_cn": 0.07046777069568634,
-   "total": 0.9584749519824982,
    "iter": 2000,
    "lr": 8.428469557121958e-05,
-   "elapsed_s": 2011.636960029602
   },
   {
-   "ce_cell": 0.17340840995311738,
-   "dice_cell": 0.3878352814912796,
-   "ce_nos": 0.03318483327515423,
-   "dice_nos": 0.2580646830797195,
-   "bce_ab": 0.029467624574899674,
-   "bce_cn": 0.07082617726176978,
-   "total": 0.9527869963645935,
    "iter": 2050,
    "lr": 7.735007872871564e-05,
-   "elapsed_s": 2061.974331855774
   },
   {
-   "ce_cell": 0.17320298582315444,
-   "dice_cell": 0.3853736054897308,
-   "ce_nos": 0.028748751198872925,
-   "dice_nos": 0.26544315218925474,
-   "bce_ab": 0.033497555926442144,
-   "bce_cn": 0.06046421557664871,
-   "total": 0.9467302620410919,
    "iter": 2100,
    "lr": 7.063735613961179e-05,
-   "elapsed_s": 2112.548271894455
   },
   {
-   "ce_cell": 0.1658191041648388,
-   "dice_cell": 0.3796718597412109,
-   "ce_nos": 0.03187917632050812,
-   "dice_nos": 0.26365615785121915,
-   "bce_ab": 0.030561734307557344,
-   "bce_cn": 0.05981527805328369,
-   "total": 0.9314033150672912,
    "iter": 2150,
    "lr": 6.416621737584896e-05,
-   "elapsed_s": 2162.667312860489
   },
   {
-   "ce_cell": 0.1701272311806679,
-   "dice_cell": 0.37863297820091246,
-   "ce_nos": 0.027262243572622538,
-   "dice_nos": 0.26668123960494994,
-   "bce_ab": 0.04351606283336878,
-   "bce_cn": 0.06579859033226967,
-   "total": 0.9520183408260345,
    "iter": 2200,
    "lr": 5.795564340238374e-05,
-   "elapsed_s": 2213.4271240234375
   },
   {
-   "ce_cell": 0.1703978794813156,
-   "dice_cell": 0.38433976292610167,
-   "ce_nos": 0.03619127416983247,
-   "dice_nos": 0.24586948692798616,
-   "bce_ab": 0.02381640898063779,
-   "bce_cn": 0.06208158168941736,
-   "total": 0.922696396112442,
    "iter": 2250,
    "lr": 5.202385090274833e-05,
-   "elapsed_s": 2263.827299118042
   },
   {
-   "ce_cell": 0.16755644142627715,
-   "dice_cell": 0.3778986144065857,
-   "ce_nos": 0.03260435258969665,
-   "dice_nos": 0.25375066459178924,
-   "bce_ab": 0.029958658767864108,
-   "bce_cn": 0.05888926636427641,
-   "total": 0.9206579995155334,
    "iter": 2300,
    "lr": 4.638823884637875e-05,
-   "elapsed_s": 2313.582683801651
   },
   {
-   "ce_cell": 0.1692182320356369,
-   "dice_cell": 0.3797942841053009,
-   "ce_nos": 0.031096484493464232,
-   "dice_nos": 0.25585519909858706,
-   "bce_ab": 0.025209803096950055,
-   "bce_cn": 0.06410345062613487,
-   "total": 0.925277452468872,
    "iter": 2350,
    "lr": 4.106533745443955e-05,
-   "elapsed_s": 2366.780472755432
   },
   {
-   "ce_cell": 0.1647430169582367,
-   "dice_cell": 0.37565741121768953,
-   "ce_nos": 0.029058436192572118,
-   "dice_nos": 0.2536770069599152,
-   "bce_ab": 0.026647686092182994,
-   "bce_cn": 0.05636850781738758,
-   "total": 0.9061520624160767,
    "iter": 2400,
    "lr": 3.607075971383718e-05,
-   "elapsed_s": 2420.7595388889313
   },
   {
-   "ce_cell": 0.1587481266260147,
-   "dice_cell": 0.37386213421821596,
-   "ce_nos": 0.027811604756861924,
-   "dice_nos": 0.2525290942192078,
-   "bce_ab": 0.027055934872478246,
-   "bce_cn": 0.05388368744403124,
-   "total": 0.8938905835151673,
    "iter": 2450,
    "lr": 3.1419155581639394e-05,
-   "elapsed_s": 2472.9100120067596
   },
   {
-   "ce_cell": 0.16981444805860518,
-   "dice_cell": 0.3800900149345398,
-   "ce_nos": 0.02935561618767679,
-   "dice_nos": 0.2700779044628143,
-   "bce_ab": 0.026528375521302223,
-   "bce_cn": 0.05692515041679144,
-   "total": 0.932791496515274,
    "iter": 2500,
    "lr": 2.7124169014226923e-05,
-   "elapsed_s": 2525.758083820343
   },
   {
-   "ce_cell": 0.16598297476768495,
-   "dice_cell": 0.37718391716480254,
-   "ce_nos": 0.030217354064807297,
-   "dice_nos": 0.2526162987947464,
-   "bce_ab": 0.027398303812369705,
-   "bce_cn": 0.05613136574625969,
-   "total": 0.9095302104949952,
    "iter": 2550,
    "lr": 2.319839794721846e-05,
-   "elapsed_s": 2578.604254961014
   },
   {
-   "ce_cell": 0.16501545295119285,
-   "dice_cell": 0.37883709609508515,
-   "ce_nos": 0.029584614327177405,
-   "dice_nos": 0.24774192690849303,
-   "bce_ab": 0.022402014611288906,
-   "bce_cn": 0.057471454478800296,
-   "total": 0.9010525631904602,
    "iter": 2600,
    "lr": 1.9653357343555055e-05,
-   "elapsed_s": 2630.874783039093
   },
   {
-   "ce_cell": 0.16610188841819762,
-   "dice_cell": 0.3750225841999054,
-   "ce_nos": 0.03136977587360889,
-   "dice_nos": 0.24075563967227936,
-   "bce_ab": 0.023018423346802593,
-   "bce_cn": 0.054641781151294706,
-   "total": 0.8909100878238678,
    "iter": 2650,
    "lr": 1.6499445418129828e-05,
-   "elapsed_s": 2683.160297870636
   },
   {
-   "ce_cell": 0.16699187859892844,
-   "dice_cell": 0.37973670542240145,
-   "ce_nos": 0.031007613111287354,
-   "dice_nos": 0.25119431257247926,
-   "bce_ab": 0.02384809180162847,
-   "bce_cn": 0.05100016433745622,
-   "total": 0.9037787771224975,
    "iter": 2700,
    "lr": 1.3745913138031944e-05,
-   "elapsed_s": 2734.7909269332886
   },
   {
-   "ce_cell": 0.1654239809513092,
-   "dice_cell": 0.373527038693428,
-   "ce_nos": 0.026881940369494258,
-   "dice_nos": 0.25084660708904266,
-   "bce_ab": 0.03065326821990311,
-   "bce_cn": 0.05875104777514935,
-   "total": 0.9060838842391967,
    "iter": 2750,
    "lr": 1.1400837087866401e-05,
-   "elapsed_s": 2785.7226259708405
   },
   {
-   "ce_cell": 0.164136121571064,
-   "dice_cell": 0.37796851992607117,
-   "ce_nos": 0.03165758168324828,
-   "dice_nos": 0.2464037835597992,
-   "bce_ab": 0.02406527513638139,
-   "bce_cn": 0.05659090880304575,
-   "total": 0.9008221983909607,
    "iter": 2800,
    "lr": 9.471095779739621e-06,
-   "elapsed_s": 2837.11731505394
   },
   {
-   "ce_cell": 0.16365474253892898,
-   "dice_cell": 0.3711686569452286,
-   "ce_nos": 0.03501185901463032,
-   "dice_nos": 0.26115318775177004,
-   "bce_ab": 0.020862439805641772,
-   "bce_cn": 0.06006271541118622,
-   "total": 0.9119135963916779,
    "iter": 2850,
    "lr": 7.96234947739845e-06,
-   "elapsed_s": 2888.5047039985657
   },
   {
-   "ce_cell": 0.1650400999188423,
-   "dice_cell": 0.37380745112895963,
-   "ce_nos": 0.027719355495646594,
-   "dice_nos": 0.24648674428462983,
-   "bce_ab": 0.030483954632654787,
-   "bce_cn": 0.05518790137022734,
-   "total": 0.898725507259369,
    "iter": 2900,
    "lr": 6.8790235937012375e-06,
-   "elapsed_s": 2939.583750963211
   },
   {
-   "ce_cell": 0.1652056121826172,
-   "dice_cell": 0.37707124650478363,
-   "ce_nos": 0.027215274684131145,
-   "dice_nos": 0.24094445288181304,
-   "bce_ab": 0.02379746310412884,
-   "bce_cn": 0.05311750005930662,
-   "total": 0.8873515450954437,
    "iter": 2950,
    "lr": 6.224295710119515e-06,
-   "elapsed_s": 2990.7978727817535
   },
   {
-   "ce_cell": 0.15959388136863709,
-   "dice_cell": 0.3714492267370224,
-   "ce_nos": 0.030986426128074528,
-   "dice_nos": 0.2347310698032379,
-   "bce_ab": 0.018123713415116073,
-   "bce_cn": 0.05247432488948107,
-   "total": 0.8673586428165436,
    "iter": 3000,
    "lr": 6.000086256343946e-06,
-   "elapsed_s": 3042.6654930114746
   }
  ],
  "val": []

 {
  "train": [
   {
+   "ce_cell": 0.9777789008617401,
+   "dice_cell": 0.8740915787220002,
+   "ce_nos": 1.036865803003311,
+   "dice_nos": 0.9926403152942658,
+   "bce_ab": 0.6191497492790222,
+   "bce_cn": 0.6784366118907929,
+   "mse_dist": 0.17519462607800962,
+   "total": 6.05493613243103,
    "iter": 50,
    "lr": 0.00015,
+   "elapsed_s": 26.069042921066284
   },
   {
+   "ce_cell": 0.7761036133766175,
+   "dice_cell": 0.8330331027507782,
+   "ce_nos": 0.7418340992927551,
+   "dice_nos": 0.9909472858905792,
+   "bce_ab": 0.4780848079919815,
+   "bce_cn": 0.4427151954174042,
+   "mse_dist": 0.028857542239129542,
+   "total": 4.407005853652954,
    "iter": 100,
    "lr": 0.0003,
+   "elapsed_s": 51.948261976242065
   },
   {
+   "ce_cell": 0.6085621345043183,
+   "dice_cell": 0.77956205368042,
+   "ce_nos": 0.5155557030439377,
+   "dice_nos": 0.9848698270320893,
+   "bce_ab": 0.38647628366947173,
+   "bce_cn": 0.3627579927444458,
+   "mse_dist": 0.022874937355518342,
+   "total": 3.752158718109131,
    "iter": 150,
    "lr": 0.00029979294712266676,
+   "elapsed_s": 77.92088508605957
   },
   {
+   "ce_cell": 0.4677648037672043,
+   "dice_cell": 0.7135179710388183,
+   "ce_nos": 0.38082618355751036,
+   "dice_nos": 0.9798667204380035,
+   "bce_ab": 0.3422890931367874,
+   "bce_cn": 0.3326674288511276,
+   "mse_dist": 0.021435761135071518,
+   "total": 3.324111022949219,
    "iter": 200,
    "lr": 0.0002991554114956718,
+   "elapsed_s": 103.77014684677124
   },
   {
+   "ce_cell": 0.38509724378585813,
+   "dice_cell": 0.653343700170517,
+   "ce_nos": 0.28423633813858035,
+   "dice_nos": 0.9686614489555359,
+   "bce_ab": 0.2873850876092911,
+   "bce_cn": 0.313295716047287,
+   "mse_dist": 0.02068787667900324,
+   "total": 2.995458931922913,
    "iter": 250,
    "lr": 0.00029808917686451123,
+   "elapsed_s": 129.53510093688965
   },
   {
+   "ce_cell": 0.33268820941448213,
+   "dice_cell": 0.5951491236686707,
+   "ce_nos": 0.21769435673952103,
+   "dice_nos": 0.9576204478740692,
+   "bce_ab": 0.2672469413280487,
+   "bce_cn": 0.2786091101169586,
+   "mse_dist": 0.020070483386516572,
+   "total": 2.7493606185913086,
    "iter": 300,
    "lr": 0.00029659737067906796,
+   "elapsed_s": 155.210942029953
   },
   {
+   "ce_cell": 0.2987843030691147,
+   "dice_cell": 0.5483147490024567,
+   "ce_nos": 0.1640540537238121,
+   "dice_nos": 0.9450577974319458,
+   "bce_ab": 0.24849256694316865,
+   "bce_cn": 0.26394254505634307,
+   "mse_dist": 0.018844751380383967,
+   "total": 2.5628697538375853,
    "iter": 350,
    "lr": 0.00029468436866391876,
+   "elapsed_s": 180.83402609825134
   },
   {
+   "ce_cell": 0.2770148387551308,
+   "dice_cell": 0.517680995464325,
+   "ce_nos": 0.13154282703995704,
+   "dice_nos": 0.9315742325782775,
+   "bce_ab": 0.22211319893598558,
+   "bce_cn": 0.26095826357603075,
+   "mse_dist": 0.018346485309302808,
+   "total": 2.4326167488098145,
    "iter": 400,
    "lr": 0.0002923557819835801,
+   "elapsed_s": 206.35285687446594
   },
   {
+   "ce_cell": 0.26904626816511157,
+   "dice_cell": 0.4866592735052109,
+   "ce_nos": 0.10414468422532082,
+   "dice_nos": 0.8954603922367096,
+   "bce_ab": 0.21155467212200166,
+   "bce_cn": 0.23948282122612,
+   "mse_dist": 0.018231311477720738,
+   "total": 2.2975046682357787,
    "iter": 450,
    "lr": 0.00028961844078399557,
+   "elapsed_s": 231.1181778907776
   },
   {
+   "ce_cell": 0.2626450803875923,
+   "dice_cell": 0.4800322437286377,
+   "ce_nos": 0.08362166196107865,
+   "dice_nos": 0.8580216205120087,
+   "bce_ab": 0.184051346629858,
+   "bce_cn": 0.24237413197755814,
+   "mse_dist": 0.018717802539467813,
+   "total": 2.204335107803345,
    "iter": 500,
    "lr": 0.0002864803741585413,
+   "elapsed_s": 256.2003290653229
   },
   {
+   "ce_cell": 0.2437168264389038,
+   "dice_cell": 0.44185430943965914,
+   "ce_nos": 0.06530485145747661,
+   "dice_nos": 0.8218306469917297,
+   "bce_ab": 0.17754501536488532,
+   "bce_cn": 0.22136381536722183,
+   "mse_dist": 0.01670815333724022,
+   "total": 2.0551562237739565,
    "iter": 550,
    "lr": 0.0002829507865973116,
+   "elapsed_s": 281.3221549987793
   },
   {
+   "ce_cell": 0.244499791264534,
+   "dice_cell": 0.43603491365909575,
+   "ce_nos": 0.056205639988183974,
+   "dice_nos": 0.764169396162033,
+   "bce_ab": 0.16857516303658485,
+   "bce_cn": 0.22340300798416138,
+   "mse_dist": 0.016940609198063613,
+   "total": 1.9775909566879273,
    "iter": 600,
    "lr": 0.0002790400309887635,
+   "elapsed_s": 306.70035195350647
   },
   {
+   "ce_cell": 0.2412768453359604,
+   "dice_cell": 0.42379125833511355,
+   "ce_nos": 0.045140089876949785,
+   "dice_nos": 0.6991792964935303,
+   "bce_ab": 0.14149282865226268,
+   "bce_cn": 0.1941063818335533,
+   "mse_dist": 0.0163105876930058,
+   "total": 1.8265396285057067,
    "iter": 650,
    "lr": 0.0002747595782529114,
+   "elapsed_s": 331.7218248844147
   },
   {
+   "ce_cell": 0.23859565168619157,
+   "dice_cell": 0.416628275513649,
+   "ce_nos": 0.03758059218525887,
+   "dice_nos": 0.6752322578430175,
+   "bce_ab": 0.14357838369905948,
+   "bce_cn": 0.19971576437354088,
+   "mse_dist": 0.016762549821287394,
+   "total": 1.7951436424255371,
    "iter": 700,
    "lr": 0.0002701219836951418,
+   "elapsed_s": 357.79720091819763
   },
   {
+   "ce_cell": 0.23537835448980332,
+   "dice_cell": 0.407176011800766,
+   "ce_nos": 0.037268193271011114,
+   "dice_nos": 0.5960061675310135,
+   "bce_ab": 0.13629250541329385,
+   "bce_cn": 0.19288790062069894,
+   "mse_dist": 0.016352202873677016,
+   "total": 1.6867701506614685,
    "iter": 750,
    "lr": 0.0002651408501793399,
+   "elapsed_s": 383.76512479782104
   },
   {
+   "ce_cell": 0.22146987557411194,
+   "dice_cell": 0.3943329083919525,
+   "ce_nos": 0.03001030534505844,
+   "dice_nos": 0.5577754044532776,
+   "bce_ab": 0.12049378417432308,
+   "bce_cn": 0.17776169538497924,
+   "mse_dist": 0.015104550141841174,
+   "total": 1.5773667359352113,
    "iter": 800,
    "lr": 0.0002598307882283455,
+   "elapsed_s": 409.39320611953735
   },
   {
+   "ce_cell": 0.2323348864912987,
+   "dice_cell": 0.40128552079200747,
+   "ce_nos": 0.033332938868552445,
+   "dice_nos": 0.5035274916887283,
+   "bce_ab": 0.10335498113185167,
+   "bce_cn": 0.17933884173631667,
+   "mse_dist": 0.015931979529559614,
+   "total": 1.5328345370292664,
    "iter": 850,
    "lr": 0.00025420737316877263,
+   "elapsed_s": 435.14874291419983
   },
   {
+   "ce_cell": 0.2182929402589798,
+   "dice_cell": 0.3786925607919693,
+   "ce_nos": 0.027653501816093922,
+   "dice_nos": 0.4839075034856796,
+   "bce_ab": 0.11399138215929269,
+   "bce_cn": 0.14645383700728418,
+   "mse_dist": 0.014609130527824164,
+   "total": 1.442037374973297,
    "iter": 900,
    "lr": 0.00024828709944589296,
+   "elapsed_s": 460.9140498638153
   },
   {
+   "ce_cell": 0.22527328491210938,
+   "dice_cell": 0.38350287079811096,
+   "ce_nos": 0.03021520377136767,
+   "dice_nos": 0.4358292263746262,
+   "bce_ab": 0.09423262968659402,
+   "bce_cn": 0.16419889584183692,
+   "mse_dist": 0.015031665433198213,
+   "total": 1.408410460948944,
    "iter": 950,
    "lr": 0.00024208733224258537,
+   "elapsed_s": 486.0358588695526
   },
   {
+   "ce_cell": 0.22444447070360185,
+   "dice_cell": 0.38110810697078706,
+   "ce_nos": 0.03478069136850536,
+   "dice_nos": 0.41253172159194945,
+   "bce_ab": 0.09447835937142372,
+   "bce_cn": 0.1629086424410343,
+   "mse_dist": 0.01506655540317297,
+   "total": 1.3855847644805908,
    "iter": 1000,
    "lr": 0.0002356262565442632,
+   "elapsed_s": 511.45353078842163
   },
   {
+   "ce_cell": 0.22289368331432344,
+   "dice_cell": 0.37346395969390866,
+   "ce_nos": 0.03566695312969387,
+   "dice_nos": 0.38205474078655244,
+   "bce_ab": 0.09222890179604291,
+   "bce_cn": 0.15356166876852512,
+   "mse_dist": 0.0148412006162107,
+   "total": 1.3340759134292604,
    "iter": 1050,
    "lr": 0.0002289228237991784,
+   "elapsed_s": 536.3748590946198
   },
   {
+   "ce_cell": 0.21837765246629715,
+   "dice_cell": 0.36276875555515287,
+   "ce_nos": 0.0334074149094522,
+   "dice_nos": 0.37752662658691405,
+   "bce_ab": 0.09045730914920569,
+   "bce_cn": 0.1526588149368763,
+   "mse_dist": 0.014266370516270398,
+   "total": 1.3065284299850464,
    "iter": 1100,
    "lr": 0.0002219966963305587,
+   "elapsed_s": 561.2481529712677
   },
   {
+   "ce_cell": 0.22264640301465988,
+   "dice_cell": 0.3701382452249527,
+   "ce_nos": 0.03397220085840672,
+   "dice_nos": 0.37801525712013245,
+   "bce_ab": 0.08047693535685539,
+   "bce_cn": 0.1416425648331642,
+   "mse_dist": 0.014651054423302412,
+   "total": 1.3001468753814698,
    "iter": 1150,
    "lr": 0.00021486818966362585,
+   "elapsed_s": 586.2862780094147
   },
   {
+   "ce_cell": 0.21802948534488678,
+   "dice_cell": 0.36143216729164124,
+   "ce_nos": 0.03354042544960976,
+   "dice_nos": 0.35177252113819124,
+   "bce_ab": 0.09707722380757332,
+   "bce_cn": 0.13814265102148057,
+   "mse_dist": 0.014300682041794062,
+   "total": 1.2714978766441345,
    "iter": 1200,
    "lr": 0.00020755821293665946,
+   "elapsed_s": 611.6254408359528
   },
   {
+   "ce_cell": 0.2118623086810112,
+   "dice_cell": 0.3594000792503357,
+   "ce_nos": 0.028813177780248226,
+   "dice_nos": 0.3532944941520691,
+   "bce_ab": 0.08322387758642436,
+   "bce_cn": 0.12183172032237052,
+   "mse_dist": 0.014152146968990564,
+   "total": 1.229186396598816,
    "iter": 1250,
    "lr": 0.00020008820757089248,
+   "elapsed_s": 636.6162917613983
   },
   {
+   "ce_cell": 0.21404922842979432,
+   "dice_cell": 0.3589194703102112,
+   "ce_nos": 0.03624305865727365,
+   "dice_nos": 0.32102711260318756,
+   "bce_ab": 0.08243383184075355,
+   "bce_cn": 0.12266159057617188,
+   "mse_dist": 0.013961246944963932,
+   "total": 1.2051405358314513,
    "iter": 1300,
    "lr": 0.0001924800843791277,
+   "elapsed_s": 661.7505779266357
   },
   {
+   "ce_cell": 0.2120766496658325,
+   "dice_cell": 0.3596819120645523,
+   "ce_nos": 0.033906866414472464,
+   "dice_nos": 0.3169685709476471,
+   "bce_ab": 0.05813073594123125,
+   "bce_cn": 0.11684628672897816,
+   "mse_dist": 0.01383887032046914,
+   "total": 1.1668053698539733,
    "iter": 1350,
    "lr": 0.00018475615929754922,
+   "elapsed_s": 686.766040802002
   },
   {
+   "ce_cell": 0.21282824724912644,
+   "dice_cell": 0.35991409242153166,
+   "ce_nos": 0.03106280479580164,
+   "dice_nos": 0.3374244701862335,
+   "bce_ab": 0.06968416810035706,
+   "bce_cn": 0.12222521275281906,
+   "mse_dist": 0.013985643181949853,
+   "total": 1.2030672073364257,
    "iter": 1400,
    "lr": 0.0001769390879292361,
+   "elapsed_s": 711.7308628559113
   },
   {
+   "ce_cell": 0.20805930137634276,
+   "dice_cell": 0.34907511532306673,
+   "ce_nos": 0.03580653930082917,
+   "dice_nos": 0.32342082798480987,
+   "bce_ab": 0.0811778188869357,
+   "bce_cn": 0.11206604793667793,
+   "mse_dist": 0.013074529767036438,
+   "total": 1.174978291988373,
    "iter": 1450,
    "lr": 0.00016905179909137546,
+   "elapsed_s": 736.551778793335
   },
   {
+   "ce_cell": 0.2046449926495552,
+   "dice_cell": 0.3410522907972336,
+   "ce_nos": 0.02621271908748895,
+   "dice_nos": 0.31531415462493895,
+   "bce_ab": 0.07398473702371121,
+   "bce_cn": 0.09817121870815754,
+   "mse_dist": 0.012903542686253787,
+   "total": 1.1238978099822998,
    "iter": 1500,
    "lr": 0.00016111742756109098,
+   "elapsed_s": 761.4780099391937
   },
   {
+   "ce_cell": 0.20931495547294618,
+   "dice_cell": 0.35134742259979246,
+   "ce_nos": 0.037548264022916555,
+   "dice_nos": 0.2945420718193054,
+   "bce_ab": 0.060994742810726164,
+   "bce_cn": 0.11476682797074318,
+   "mse_dist": 0.013118454609066249,
+   "total": 1.134106569290161,
    "iter": 1550,
    "lr": 0.00015315924621715527,
+   "elapsed_s": 786.305233001709
   },
   {
+   "ce_cell": 0.20947466552257538,
+   "dice_cell": 0.34759899854660037,
+   "ce_nos": 0.03775742853060365,
+   "dice_nos": 0.2938369768857956,
+   "bce_ab": 0.05887895548716188,
+   "bce_cn": 0.1020959459990263,
+   "mse_dist": 0.013232378214597702,
+   "total": 1.1158048474788667,
    "iter": 1600,
    "lr": 0.00014520059777662684,
+   "elapsed_s": 811.053496837616
   },
   {
+   "ce_cell": 0.20545023560523987,
+   "dice_cell": 0.34604851961135863,
+   "ce_nos": 0.03138292274437845,
+   "dice_nos": 0.3009911799430847,
+   "bce_ab": 0.051527918446809054,
+   "bce_cn": 0.09074965998530388,
+   "mse_dist": 0.012758474349975586,
+   "total": 1.0899428081512452,
    "iter": 1650,
    "lr": 0.00013726482632663912,
+   "elapsed_s": 835.8836717605591
   },
   {
+   "ce_cell": 0.2035794249176979,
+   "dice_cell": 0.3380417960882187,
+   "ce_nos": 0.03119728611782193,
+   "dice_nos": 0.29003060042858125,
+   "bce_ab": 0.055323713328689336,
+   "bce_cn": 0.0933622520416975,
+   "mse_dist": 0.012566495817154646,
+   "total": 1.0743675696849824,
    "iter": 1700,
    "lr": 0.00012937520885217038,
+   "elapsed_s": 860.7524697780609
   },
   {
+   "ce_cell": 0.2061012852191925,
+   "dice_cell": 0.34122552573680875,
+   "ce_nos": 0.03216092755552381,
+   "dice_nos": 0.2873716199398041,
+   "bce_ab": 0.042616604026407004,
+   "bce_cn": 0.09780772387981415,
+   "mse_dist": 0.013027021698653697,
+   "total": 1.072418793439865,
    "iter": 1750,
    "lr": 0.0001215548869606366,
+   "elapsed_s": 885.5479848384857
   },
   {
+   "ce_cell": 0.2040314009785652,
+   "dice_cell": 0.34347267031669615,
+   "ce_nos": 0.031174597078934312,
+   "dice_nos": 0.2863872331380844,
+   "bce_ab": 0.05184935016557574,
+   "bce_cn": 0.08530131049454212,
+   "mse_dist": 0.012656427212059498,
+   "total": 1.0654987025260925,
    "iter": 1800,
    "lr": 0.00011382679900356957,
+   "elapsed_s": 910.4776349067688
   },
   {
+   "ce_cell": 0.19866832792758943,
+   "dice_cell": 0.3420099478960037,
+   "ce_nos": 0.03141806117258966,
+   "dice_nos": 0.27803765118122103,
+   "bce_ab": 0.045098451543599365,
+   "bce_cn": 0.09952721245586872,
+   "mse_dist": 0.012453611139208079,
+   "total": 1.0570277082920074,
    "iter": 1850,
    "lr": 0.0001062136127944796,
+   "elapsed_s": 935.2265169620514
   },
   {
+   "ce_cell": 0.19916877180337905,
+   "dice_cell": 0.34010264158248904,
+   "ce_nos": 0.030681413793936373,
+   "dice_nos": 0.268166041970253,
+   "bce_ab": 0.04972605316899717,
+   "bce_cn": 0.08704228073358536,
+   "mse_dist": 0.012498341277241707,
+   "total": 1.0373789024353028,
    "iter": 1900,
    "lr": 9.873765912025276e-05,
+   "elapsed_s": 960.0961599349976
   },
   {
+   "ce_cell": 0.19386377602815627,
+   "dice_cell": 0.334270116686821,
+   "ce_nos": 0.02582207920961082,
+   "dice_nos": 0.2805849975347519,
+   "bce_ab": 0.05429890226572752,
+   "bce_cn": 0.07757438875734807,
+   "mse_dist": 0.012007349245250225,
+   "total": 1.026451007127762,
    "iter": 1950,
    "lr": 9.142086624110557e-05,
+   "elapsed_s": 985.0865857601166
   },
   {
+   "ce_cell": 0.2070612996816635,
+   "dice_cell": 0.3450821244716644,
+   "ce_nos": 0.03970785442739725,
+   "dice_nos": 0.27347362697124483,
+   "bce_ab": 0.04065307324752212,
+   "bce_cn": 0.09133856296539307,
+   "mse_dist": 0.012714939899742603,
+   "total": 1.0608912444114684,
    "iter": 2000,
    "lr": 8.428469557121958e-05,
+   "elapsed_s": 1010.4481308460236
   },
   {
+   "ce_cell": 0.20134998142719268,
+   "dice_cell": 0.33819403886795046,
+   "ce_nos": 0.030352213024161755,
+   "dice_nos": 0.27864284574985504,
+   "bce_ab": 0.049128259383142,
+   "bce_cn": 0.08518408581614495,
+   "mse_dist": 0.012358254436403513,
+   "total": 1.044642698764801,
    "iter": 2050,
    "lr": 7.735007872871564e-05,
+   "elapsed_s": 1035.885332107544
   },
   {
+   "ce_cell": 0.19757086873054505,
+   "dice_cell": 0.33871914267539976,
+   "ce_nos": 0.02972847773693502,
+   "dice_nos": 0.2694511997699738,
+   "bce_ab": 0.04641149152070284,
+   "bce_cn": 0.07878688346594571,
+   "mse_dist": 0.012009835205972194,
+   "total": 1.0207172453403472,
    "iter": 2100,
    "lr": 7.063735613961179e-05,
+   "elapsed_s": 1061.2731919288635
   },
   {
+   "ce_cell": 0.1988272961974144,
+   "dice_cell": 0.33160640120506285,
+   "ce_nos": 0.03366862212307751,
+   "dice_nos": 0.2641986829042435,
+   "bce_ab": 0.04372629556804895,
+   "bce_cn": 0.08536963656544685,
+   "mse_dist": 0.011948191411793232,
+   "total": 1.0171378910541535,
    "iter": 2150,
    "lr": 6.416621737584896e-05,
+   "elapsed_s": 1086.707243680954
   },
   {
+   "ce_cell": 0.19380151480436325,
+   "dice_cell": 0.32752876222133637,
+   "ce_nos": 0.03283986233174801,
+   "dice_nos": 0.26049777925014495,
+   "bce_ab": 0.041361887510865926,
+   "bce_cn": 0.07517353564500809,
+   "mse_dist": 0.011463552955538034,
+   "total": 0.9885211074352265,
    "iter": 2200,
    "lr": 5.795564340238374e-05,
+   "elapsed_s": 1112.0149159431458
   },
   {
+   "ce_cell": 0.19721020132303238,
+   "dice_cell": 0.33173409938812254,
+   "ce_nos": 0.02758968804962933,
+   "dice_nos": 0.27361033380031585,
+   "bce_ab": 0.04173741614446044,
+   "bce_cn": 0.07369629118591547,
+   "mse_dist": 0.01197859773412347,
+   "total": 1.0054710137844085,
    "iter": 2250,
    "lr": 5.202385090274833e-05,
+   "elapsed_s": 1137.4904730319977
   },
   {
+   "ce_cell": 0.18801554799079895,
+   "dice_cell": 0.32500296354293823,
+   "ce_nos": 0.026653141435235738,
+   "dice_nos": 0.2547152751684189,
+   "bce_ab": 0.03614282112568617,
+   "bce_cn": 0.07250596221536398,
+   "mse_dist": 0.011296509951353072,
+   "total": 0.9595182621479035,
    "iter": 2300,
    "lr": 4.638823884637875e-05,
+   "elapsed_s": 1162.9490859508514
   },
   {
+   "ce_cell": 0.19445410251617432,
+   "dice_cell": 0.3265311974287033,
+   "ce_nos": 0.02529556589666754,
+   "dice_nos": 0.26673332929611204,
+   "bce_ab": 0.05173649054020643,
+   "bce_cn": 0.07536047860980034,
+   "mse_dist": 0.011767324544489383,
+   "total": 0.9989477896690369,
    "iter": 2350,
    "lr": 4.106533745443955e-05,
+   "elapsed_s": 1188.4017899036407
   },
   {
+   "ce_cell": 0.19469613671302796,
+   "dice_cell": 0.33241518437862394,
+   "ce_nos": 0.033744015344418586,
+   "dice_nos": 0.24858344793319703,
+   "bce_ab": 0.03829080337658525,
+   "bce_cn": 0.07345383018255233,
+   "mse_dist": 0.011775377187877894,
+   "total": 0.980060316324234,
    "iter": 2400,
    "lr": 3.607075971383718e-05,
+   "elapsed_s": 1214.008955001831
   },
   {
+   "ce_cell": 0.19300242602825166,
+   "dice_cell": 0.32622973263263705,
+   "ce_nos": 0.023967032334767283,
+   "dice_nos": 0.2765471911430359,
+   "bce_ab": 0.04265340869314969,
+   "bce_cn": 0.07023378118872642,
+   "mse_dist": 0.01163030793890357,
+   "total": 0.9907851052284241,
    "iter": 2450,
    "lr": 3.1419155581639394e-05,
+   "elapsed_s": 1238.943860054016
   },
   {
+   "ce_cell": 0.19708714127540589,
+   "dice_cell": 0.3298483777046204,
+   "ce_nos": 0.035931861903518436,
+   "dice_nos": 0.2631306803226471,
+   "bce_ab": 0.03939212892204523,
+   "bce_cn": 0.07550869181752205,
+   "mse_dist": 0.011621671505272388,
+   "total": 0.99900723695755,
    "iter": 2500,
    "lr": 2.7124169014226923e-05,
+   "elapsed_s": 1264.1307349205017
   },
   {
+   "ce_cell": 0.19627166271209717,
+   "dice_cell": 0.3276835995912552,
+   "ce_nos": 0.030831458177417517,
+   "dice_nos": 0.2647592234611511,
+   "bce_ab": 0.035610632617026565,
+   "bce_cn": 0.06968229006975889,
+   "mse_dist": 0.011490394324064256,
+   "total": 0.98229083776474,
    "iter": 2550,
    "lr": 2.319839794721846e-05,
+   "elapsed_s": 1288.9724678993225
   },
   {
+   "ce_cell": 0.19514285027980804,
+   "dice_cell": 0.3292509812116623,
+   "ce_nos": 0.02648238055408001,
+   "dice_nos": 0.26873668015003205,
+   "bce_ab": 0.03310712873935699,
+   "bce_cn": 0.06389056332409382,
+   "mse_dist": 0.01165549136698246,
+   "total": 0.9748880290985107,
    "iter": 2600,
    "lr": 1.9653357343555055e-05,
+   "elapsed_s": 1313.9768908023834
   },
   {
+   "ce_cell": 0.19508351147174835,
+   "dice_cell": 0.32727956473827363,
+   "ce_nos": 0.0280744384881109,
+   "dice_nos": 0.26588015496730805,
+   "bce_ab": 0.0365638392791152,
+   "bce_cn": 0.07343160055577755,
+   "mse_dist": 0.011491857282817364,
+   "total": 0.9837723982334137,
    "iter": 2650,
    "lr": 1.6499445418129828e-05,
+   "elapsed_s": 1338.650880098343
   },
   {
+   "ce_cell": 0.19099752217531205,
+   "dice_cell": 0.32075556099414826,
+   "ce_nos": 0.029300750745460392,
+   "dice_nos": 0.2554633593559265,
+   "bce_ab": 0.045620939303189514,
+   "bce_cn": 0.07410217750817537,
+   "mse_dist": 0.01112760055810213,
+   "total": 0.971878319978714,
    "iter": 2700,
    "lr": 1.3745913138031944e-05,
+   "elapsed_s": 1363.5123808383942
   },
   {
+   "ce_cell": 0.1883079969882965,
+   "dice_cell": 0.32481179118156434,
+   "ce_nos": 0.031208774596452712,
+   "dice_nos": 0.2581752932071686,
+   "bce_ab": 0.038093772381544114,
+   "bce_cn": 0.06765574529767036,
+   "mse_dist": 0.011228110697120429,
+   "total": 0.9643939423561096,
    "iter": 2750,
    "lr": 1.1400837087866401e-05,
+   "elapsed_s": 1388.352870941162
   },
   {
+   "ce_cell": 0.1916240319609642,
+   "dice_cell": 0.326035658121109,
+   "ce_nos": 0.023255940917879345,
+   "dice_nos": 0.2671393311023712,
+   "bce_ab": 0.03772587236016989,
+   "bce_cn": 0.0679256134480238,
+   "mse_dist": 0.011575959492474794,
+   "total": 0.9715862393379211,
    "iter": 2800,
    "lr": 9.471095779739621e-06,
+   "elapsed_s": 1413.569692850113
   },
   {
+   "ce_cell": 0.19167108833789825,
+   "dice_cell": 0.32803118288516997,
+   "ce_nos": 0.02982857631519437,
+   "dice_nos": 0.25197103917598723,
+   "bce_ab": 0.03527032459154725,
+   "bce_cn": 0.06827950268983841,
+   "mse_dist": 0.011356610022485256,
+   "total": 0.961834750175476,
    "iter": 2850,
    "lr": 7.96234947739845e-06,
+   "elapsed_s": 1438.4622449874878
   },
   {
+   "ce_cell": 0.19785686254501342,
+   "dice_cell": 0.3267842435836792,
+   "ce_nos": 0.03312531067058444,
+   "dice_nos": 0.2531671738624573,
+   "bce_ab": 0.036567215006798506,
+   "bce_cn": 0.06692319795489311,
+   "mse_dist": 0.011690724939107895,
+   "total": 0.9728776228427887,
    "iter": 2900,
    "lr": 6.8790235937012375e-06,
+   "elapsed_s": 1463.5729517936707
   },
   {
+   "ce_cell": 0.1914523333311081,
+   "dice_cell": 0.3220532989501953,
+   "ce_nos": 0.02655307489912957,
+   "dice_nos": 0.26563074231147765,
+   "bce_ab": 0.04149866354651749,
+   "bce_cn": 0.06330326247960329,
+   "mse_dist": 0.011330173863098026,
+   "total": 0.9671422529220581,
    "iter": 2950,
    "lr": 6.224295710119515e-06,
+   "elapsed_s": 1481.2963027954102
   },
   {
+   "ce_cell": 0.19723920345306398,
+   "dice_cell": 0.331863129734993,
+   "ce_nos": 0.031104682530276477,
+   "dice_nos": 0.254480841755867,
+   "bce_ab": 0.036050576698035,
+   "bce_cn": 0.070169439278543,
+   "mse_dist": 0.011816464867442845,
+   "total": 0.9799901854991913,
    "iter": 3000,
    "lr": 6.000086256343946e-06,
+   "elapsed_s": 1496.7248327732086
   }
  ],
  "val": []