Parth1503 commited on
Commit
c27e84f
·
verified ·
1 Parent(s): 7b0d829

Correct Multi-Scanner SCC overlap: not name-joinable, joinable by 4um/px dimensions (44/44)

Browse files
Files changed (1) hide show
  1. README.md +17 -70
README.md CHANGED
@@ -1,76 +1,17 @@
1
  ---
2
  license: cc-by-4.0
3
  task_categories:
4
- - image-segmentation
5
  tags:
6
- - medical
7
- - histopathology
8
- - whole-slide-imaging
9
- - veterinary
10
- - canine
11
- - dermatology
12
- - oncology
13
  size_categories:
14
- - n<1K
15
- configs:
16
- - config_name: default
17
- data_files:
18
- - split: train
19
- path: data/train-*
20
- - split: val
21
- path: data/val-*
22
- - split: test
23
- path: data/test-*
24
- dataset_info:
25
- features:
26
- - name: image
27
- dtype: image
28
- - name: mask
29
- dtype: image
30
- - name: slide
31
- dtype: string
32
- - name: stem
33
- dtype: string
34
- - name: subtype
35
- dtype: string
36
- - name: patient_id
37
- dtype: string
38
- - name: split
39
- dtype: string
40
- - name: scanner
41
- dtype: string
42
- - name: tumor_class_id
43
- dtype: int32
44
- - name: tumor_class_name
45
- dtype: string
46
- - name: width
47
- dtype: int32
48
- - name: height
49
- dtype: int32
50
- - name: mpp
51
- dtype: float32
52
- - name: downsample
53
- dtype: float32
54
- - name: level0_width
55
- dtype: int32
56
- - name: level0_height
57
- dtype: int32
58
- - name: annotated_frac
59
- dtype: float32
60
- - name: classes_present
61
- dtype: string
62
- splits:
63
- - name: train
64
- num_bytes: 8696171799
65
- num_examples: 245
66
- - name: val
67
- num_bytes: 1066103799
68
- num_examples: 35
69
- - name: test
70
- num_bytes: 2288552020
71
- num_examples: 70
72
- download_size: 12051782257
73
- dataset_size: 12050827618
74
  ---
75
 
76
  # CATCH — CAnine CuTaneous Cancer Histology
@@ -184,8 +125,14 @@ much of the remainder is glass, not untraced tissue).
184
  ## Overlap with other datasets — leakage warnings
185
 
186
  - **Multi-Scanner Canine Cutaneous SCC** ([Zenodo 7418555](https://zenodo.org/records/7418555))
187
- re-scans **44 of CATCH's 50 SCC slides** on 4 additional scanners. Confirmed
188
- by the authors and **joinable by filename**. Do not mix.
 
 
 
 
 
 
189
  - **MIDOG++ / MIDOG 2022 Domain 4** is 50 canine cutaneous mast cell tumor
190
  cases from the same archive, scanner and resolution as CATCH's 50 MCT
191
  slides. The reuse is undocumented and **no cross-reference ID exists** — the
 
1
  ---
2
  license: cc-by-4.0
3
  task_categories:
4
+ - image-segmentation
5
  tags:
6
+ - medical
7
+ - histopathology
8
+ - whole-slide-imaging
9
+ - veterinary
10
+ - canine
11
+ - dermatology
12
+ - oncology
13
  size_categories:
14
+ - n<1K
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
15
  ---
16
 
17
  # CATCH — CAnine CuTaneous Cancer Histology
 
125
  ## Overlap with other datasets — leakage warnings
126
 
127
  - **Multi-Scanner Canine Cutaneous SCC** ([Zenodo 7418555](https://zenodo.org/records/7418555))
128
+ re-scans **44 of CATCH's 50 SCC slides** on 4 additional scanners (220 images,
129
+ also distributed at 4 µm/px). It is **not joinable by name** — its files are
130
+ renumbered `scc_01`…`scc_44` and the shipped COCO/SQLite carry no CATCH
131
+ provenance — but it **is joinable by 4 µm/px image dimensions**: each of its
132
+ 44 Aperio-CS2 images matches exactly one CATCH SCC slide, **44/44 with zero
133
+ ambiguity**. The six SCC slides *not* re-scanned are `SCC_08_1`, `SCC_11_1`,
134
+ `SCC_12_3`, `SCC_16_1`, `SCC_27_1`, `SCC_28_1`; exposure by split is
135
+ train 30/35, val 5/5, test 9/10, over 29 of the 33 SCC patients.
136
  - **MIDOG++ / MIDOG 2022 Domain 4** is 50 canine cutaneous mast cell tumor
137
  cases from the same archive, scanner and resolution as CATCH's 50 MCT
138
  slides. The reuse is undocumented and **no cross-reference ID exists** — the