File size: 6,255 Bytes
37e13f7 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 | """ROI border overlays for Huth story-listening flatmaps."""
from __future__ import annotations
from dataclasses import dataclass
from pathlib import Path
import h5py
import numpy as np
from PIL import Image, ImageDraw
from scipy import sparse
from scipy.ndimage import binary_closing, binary_dilation, binary_erosion
from flatmap_plotting import text_size
@dataclass(frozen=True)
class ROIGroup:
label: str
description: str
color: tuple[int, int, int]
rois: tuple[str, ...]
ROI_GROUPS = (
ROIGroup("AC", "auditory cortex", (0, 220, 255), ("AC",)),
ROIGroup("Lang-frontal", "Broca", (255, 180, 0), ("Broca",)),
ROIGroup("Lang-temporal", "STG/STS/LTC/TC", (255, 80, 0), ("STG", "STG_old", "STS", "LTC", "TC")),
ROIGroup("Lang-parietal", "LPC/cIPL", (255, 225, 0), ("LPC", "cIPL")),
ROIGroup(
"High-level visual",
"PPA/RSC/OPA/FFA/FBA/OFA/EBA/LO/VTC/hMT",
(255, 35, 210),
("PPA", "RSC", "OPA", "FFA", "FBA", "OFA", "EBA", "LO", "VTC", "hMT"),
),
ROIGroup("Dorsal attention", "IPS/FEF/SEF", (170, 105, 255), ("IPS", "FEF", "SEF")),
ROIGroup("Frontal control", "FC/FO/MPC", (65, 95, 255), ("FC", "FO", "MPC")),
)
def load_roi_group_masks(data_root: Path, subject: str, threshold: float = 0.25) -> list[dict[str, object]]:
mapper_path = data_root / "mappers" / f"{subject}_mappers.hdf"
groups = []
with h5py.File(mapper_path, "r") as h5:
shape = tuple(int(value) for value in h5["voxel_to_flatmap_shape"][:])
mapper = sparse.csr_matrix(
(
h5["voxel_to_flatmap_data"][:],
h5["voxel_to_flatmap_indices"][:],
h5["voxel_to_flatmap_indptr"][:],
),
shape=shape,
)
flatmap_mask = h5["flatmap_mask"][:].astype(bool)
denominator = np.asarray(mapper @ np.ones(shape[1], dtype=np.float32), dtype=np.float32)
active = denominator > 1e-6
for group in ROI_GROUPS:
available = [name for name in group.rois if f"roi_mask_{name}" in h5]
if not available:
continue
voxel_mask = np.zeros(shape[1], dtype=np.float32)
for name in available:
voxel_mask = np.maximum(voxel_mask, (np.asarray(h5[f"roi_mask_{name}"][:]) > 0.5).astype(np.float32))
weighted = np.asarray(mapper @ voxel_mask, dtype=np.float32)
coverage = np.zeros_like(weighted)
coverage[active] = weighted[active] / denominator[active]
flatmap_roi = np.zeros(flatmap_mask.shape, dtype=bool)
flatmap_roi[flatmap_mask] = coverage > threshold
if np.any(flatmap_roi):
groups.append(
{
"label": group.label,
"description": group.description,
"color": group.color,
"rois": available,
"mask": flatmap_roi,
}
)
return groups
def border_from_mask(mask: np.ndarray, panel_mask: np.ndarray, width: int, smooth_iterations: int) -> np.ndarray:
if not np.any(mask):
return np.zeros_like(mask, dtype=bool)
structure = np.ones((3, 3), dtype=bool)
smooth_mask = mask
if smooth_iterations > 0:
smooth_mask = binary_closing(mask, structure=structure, iterations=smooth_iterations)
border = binary_dilation(smooth_mask, structure=structure, iterations=width) & ~binary_erosion(
smooth_mask,
structure=structure,
iterations=width,
border_value=0,
)
return border & panel_mask
def overlay_roi_borders(
image: Image.Image,
roi_groups: list[dict[str, object]],
crop: tuple[slice, slice],
panel_mask: np.ndarray,
width: int = 1,
smooth_iterations: int = 1,
dot_step: int = 7,
) -> Image.Image:
if not roi_groups:
return image
rgba = np.asarray(image.convert("RGBA")).copy()
for group in roi_groups:
roi_mask = np.asarray(group["mask"], dtype=bool)[crop] & panel_mask
border = border_from_mask(roi_mask, panel_mask, width, smooth_iterations)
if not np.any(border):
continue
halo = binary_dilation(border, structure=np.ones((3, 3), dtype=bool), iterations=1) & panel_mask
rgba[halo, :3] = (20, 20, 20)
rgba[halo, 3] = 255
if dot_step > 0:
rows, cols = np.indices(roi_mask.shape)
dots = roi_mask & panel_mask & ~border & (((rows * 3 + cols * 5) % dot_step) == 0)
rgba[dots, :3] = group["color"]
rgba[dots, 3] = 255
rgba[border, :3] = group["color"]
rgba[border, 3] = 255
return Image.fromarray(rgba, mode="RGBA")
def roi_group_metadata(roi_groups: list[dict[str, object]]) -> list[dict[str, object]]:
return [
{
"label": str(group["label"]),
"description": str(group["description"]),
"color_rgb": list(group["color"]),
"rois": list(group["rois"]),
}
for group in roi_groups
]
def draw_roi_legend(
draw: ImageDraw.ImageDraw,
roi_groups: list[dict[str, object]],
x: int,
y: int,
max_width: int,
font,
title_font,
) -> int:
if not roi_groups:
return y
title = "ROI borders:"
draw.text((x, y), title, fill=(20, 20, 20), font=title_font)
title_w, _ = text_size(draw, title, title_font)
cursor_x = x + title_w + 18
cursor_y = y + 1
line_h = 24
for group in roi_groups:
label = f"{group['label']}: {group['description']}"
label_w, _ = text_size(draw, label, font)
item_w = 34 + label_w + 22
if cursor_x + item_w > x + max_width and cursor_x > x + title_w + 18:
cursor_x = x
cursor_y += line_h
color = tuple(int(value) for value in group["color"])
line_y = cursor_y + 11
draw.line((cursor_x, line_y, cursor_x + 24, line_y), fill=(0, 0, 0), width=5)
draw.line((cursor_x, line_y, cursor_x + 24, line_y), fill=color, width=3)
draw.text((cursor_x + 32, cursor_y), label, fill=(20, 20, 20), font=font)
cursor_x += item_w
return cursor_y + line_h
|