Dataset Viewer
Auto-converted to Parquet Duplicate
patch_id
int32
image
image
instance_mask
image
cancer_type
string
wsi_id
string
x
int32
y
int32
size_original
int32
size_in_40x
int32
num_instances
int32
has_multirater
bool
1
blca
TCGA-2F-A9KR-01Z-00-DX1
107,045
21,476
439
400
82
false
2
blca
TCGA-4Z-AA81-01Z-00-DX1
60,461
50,884
396
400
9
false
3
blca
TCGA-4Z-AA89-01Z-00-DX1
21,643
44,260
396
400
21
false
4
blca
TCGA-4Z-AA89-01Z-00-DX1
61,314
67,446
396
400
31
false
5
blca
TCGA-5N-A9KM-01Z-00-DX1
63,542
33,240
514
400
14
false
6
blca
TCGA-BL-A0C8-01Z-00-DX1
111,817
21,574
405
400
5
false
7
blca
TCGA-BL-A3JM-01Z-00-DX1
18,483
32,248
407
400
52
false
8
blca
TCGA-BT-A20P-01Z-00-DX1
76,444
36,570
405
400
12
false
9
blca
TCGA-BT-A2LB-01Z-00-DX1
62,714
12,465
405
400
21
false
10
blca
TCGA-BT-A42B-01Z-00-DX1
70,253
7,113
407
400
16
false
11
blca
TCGA-CF-A47T-01Z-00-DX1
75,760
47,094
403
400
51
false
12
blca
TCGA-CF-A47W-01Z-00-DX1
37,271
19,621
403
400
10
false
13
blca
TCGA-CF-A9FL-01Z-00-DX1
50,553
60,829
396
400
17
false
14
blca
TCGA-CU-A3KJ-01Z-00-DX1
95,491
25,819
404
400
21
false
15
blca
TCGA-E7-A3Y1-01Z-00-DX1
38,828
56,119
405
400
21
false
16
blca
TCGA-E7-A6ME-01Z-00-DX1
26,470
43,620
405
400
38
false
17
blca
TCGA-E7-A7XN-01Z-00-DX1
49,907
28,868
395
400
30
false
18
blca
TCGA-E7-A85H-01Z-00-DX1
102,576
17,213
396
400
90
false
19
blca
TCGA-FD-A3B6-01Z-00-DX1
67,892
16,839
405
400
10
false
20
blca
TCGA-FD-A3B6-01Z-00-DX1
75,936
31,784
405
400
11
false
21
blca
TCGA-FD-A3B8-01Z-00-DX1
41,411
52,508
405
400
17
false
22
blca
TCGA-FD-A3NA-01Z-00-DX1
28,824
28,421
405
400
21
false
23
blca
TCGA-FD-A3SM-01Z-00-DX1
66,322
21,383
402
400
23
false
24
blca
TCGA-FD-A3SN-01Z-00-DX1
25,910
54,149
402
400
24
false
25
blca
TCGA-FD-A3SN-01Z-00-DX1
97,576
18,913
402
400
6
false
26
blca
TCGA-FD-A3SO-01Z-00-DX1
95,725
49,687
402
400
3
false
27
blca
TCGA-FD-A43P-01Z-00-DX1
39,936
8,445
395
400
5
false
28
blca
TCGA-FD-A43P-01Z-00-DX1
85,017
42,318
395
400
5
false
29
blca
TCGA-FD-A43U-01Z-00-DX1
25,262
69,231
395
400
77
false
30
blca
TCGA-FD-A43X-01Z-00-DX1
82,715
28,130
395
400
40
false
31
blca
TCGA-FD-A62P-01Z-00-DX1
26,203
37,691
395
400
9
false
32
blca
TCGA-FD-A6TB-01Z-00-DX1
43,566
70,731
395
400
18
false
33
blca
TCGA-FD-A6TC-01Z-00-DX1
84,646
40,486
395
400
27
false
34
blca
TCGA-FD-A6TH-01Z-00-DX1
66,159
13,901
395
400
3
true
35
blca
TCGA-FD-A6TI-01Z-00-DX1
36,545
60,840
395
400
11
false
36
blca
TCGA-FJ-A3Z7-01Z-00-DX6
85,554
12,044
407
400
11
false
37
blca
TCGA-FT-A61P-01Z-00-DX1
33,060
26,688
402
400
31
false
38
blca
TCGA-G2-A2EC-01Z-00-DX6
36,159
48,982
403
400
1
false
39
blca
TCGA-G2-A2EC-01Z-00-DX6
53,068
47,268
403
400
39
false
40
blca
TCGA-G2-A2EJ-01Z-00-DX3
14,813
32,959
403
400
48
false
41
blca
TCGA-G2-A2EK-01Z-00-DX2
18,959
9,725
403
400
6
false
42
blca
TCGA-G2-A2EK-01Z-00-DX3
73,869
22,686
403
400
31
false
43
blca
TCGA-G2-A2EO-01Z-00-DX3
65,348
46,519
403
400
10
false
44
blca
TCGA-G2-A2EO-01Z-00-DX3
75,845
19,427
403
400
23
false
45
blca
TCGA-G2-A2EO-01Z-00-DX5
26,468
26,889
403
400
34
false
46
blca
TCGA-G2-A2EO-01Z-00-DX8
55,600
40,031
403
400
7
false
47
blca
TCGA-G2-A2ES-01Z-00-DX3
80,775
54,730
403
400
11
false
48
blca
TCGA-G2-A2ES-01Z-00-DX5
61,191
75,478
403
400
1
false
50
blca
TCGA-G2-A3IB-01Z-00-DX4
91,637
35,667
403
400
8
false
51
blca
TCGA-G2-A3IB-01Z-00-DX4
9,890
14,964
403
400
12
false
52
blca
TCGA-G2-A3IE-01Z-00-DX2
119,757
30,695
403
400
31
false
53
blca
TCGA-G2-AA3C-01Z-00-DX2
77,281
14,041
396
400
24
false
54
blca
TCGA-G2-AA3D-01Z-00-DX1
54,440
21,717
396
400
12
false
55
blca
TCGA-G2-AA3F-01Z-00-DX1
114,959
37,819
396
400
0
false
56
blca
TCGA-G2-AA3F-01Z-00-DX1
31,713
16,121
396
400
14
false
57
blca
TCGA-G2-AA3F-01Z-00-DX2
60,859
62,068
396
400
46
false
58
blca
TCGA-G2-AA3F-01Z-00-DX6
110,185
20,150
396
400
22
false
60
blca
TCGA-GC-A3YS-01Z-00-DX1
37,343
23,186
407
400
21
false
61
blca
TCGA-GC-A4ZW-01Z-00-DX1
20,673
65,157
396
400
19
false
62
blca
TCGA-GD-A3OP-01Z-00-DX1
27,751
5,387
403
400
32
false
63
blca
TCGA-GD-A3OP-01Z-00-DX1
81,781
31,575
403
400
20
false
64
blca
TCGA-GD-A3OQ-01Z-00-DX1
124,044
27,570
405
400
18
false
65
blca
TCGA-GU-A42P-01Z-00-DX1
32,602
80,316
403
400
5
false
66
blca
TCGA-GU-A42Q-01Z-00-DX1
25,855
22,222
403
400
5
false
67
blca
TCGA-GU-A42Q-01Z-00-DX1
83,897
42,549
403
400
13
false
68
blca
TCGA-GU-A42Q-01Z-00-DX2
61,434
45,209
396
400
23
true
69
blca
TCGA-GU-A763-01Z-00-DX1
23,627
13,734
396
400
1
false
70
blca
TCGA-GU-A764-01Z-00-DX1
8,913
38,724
396
400
15
false
71
blca
TCGA-GU-A766-01Z-00-DX1
76,885
44,977
396
400
11
false
72
blca
TCGA-K4-A4AC-01Z-00-DX1
23,171
52,936
407
400
36
false
73
blca
TCGA-K4-A4AC-01Z-00-DX1
33,303
61,151
407
400
21
false
74
blca
TCGA-K4-A5RI-01Z-00-DX1
163,294
51,343
402
400
17
false
75
blca
TCGA-K4-A6MB-01Z-00-DX1
61,743
56,379
402
400
6
false
76
blca
TCGA-S5-A6DX-01Z-00-DX1
21,840
37,445
396
400
23
false
77
blca
TCGA-UY-A78M-01Z-00-DX1
55,319
34,161
395
400
10
false
78
blca
TCGA-UY-A78N-01Z-00-DX1
33,696
68,244
395
400
64
false
79
blca
TCGA-UY-A78N-01Z-00-DX1
61,806
51,643
395
400
8
false
80
blca
TCGA-UY-A78N-01Z-00-DX1
68,394
49,269
395
400
19
false
81
blca
TCGA-UY-A8OB-01Z-00-DX1
113,787
18,582
395
400
14
false
82
blca
TCGA-UY-A9PB-01Z-00-DX1
31,208
67,891
395
400
29
false
83
blca
TCGA-UY-A9PH-01Z-00-DX1
45,292
32,278
395
400
37
false
84
blca
TCGA-XF-A8HI-01Z-00-DX1
79,128
7,707
395
400
11
false
85
blca
TCGA-XF-A9SG-01Z-00-DX1
35,565
14,290
395
400
10
false
86
blca
TCGA-XF-A9SK-01Z-00-DX1
32,337
33,340
395
400
25
false
87
blca
TCGA-XF-A9SL-01Z-00-DX1
96,451
83,503
395
400
4
false
88
blca
TCGA-XF-A9ST-01Z-00-DX1
15,210
47,458
395
400
77
false
89
blca
TCGA-XF-A9SV-01Z-00-DX1
121,896
8,473
395
400
10
false
90
blca
TCGA-XF-A9T3-01Z-00-DX1
54,155
57,335
395
400
15
false
91
blca
TCGA-XF-AAMG-01Z-00-DX1
19,492
51,366
395
400
6
false
92
blca
TCGA-XF-AAMQ-01Z-00-DX1
57,969
37,178
395
400
10
false
93
blca
TCGA-XF-AAMQ-01Z-00-DX1
58,644
38,801
395
400
6
false
94
blca
TCGA-XF-AAMQ-01Z-00-DX1
59,644
25,868
395
400
21
false
95
blca
TCGA-XF-AAMX-01Z-00-DX1
38,713
6,044
395
400
15
false
96
blca
TCGA-XF-AAMX-01Z-00-DX1
42,608
26,563
395
400
4
false
97
blca
TCGA-XF-AAMY-01Z-00-DX1
82,376
82,185
395
400
25
false
98
blca
TCGA-XF-AAN2-01Z-00-DX1
93,166
33,201
395
400
42
false
99
blca
TCGA-XF-AAN3-01Z-00-DX1
61,558
63,110
395
400
50
false
100
brca
TCGA-A1-A0SP-01Z-00-DX1
73,940
19,775
396
400
35
false
101
brca
TCGA-A2-A0CU-01Z-00-DX1
11,229
17,937
402
400
24
false
102
brca
TCGA-A2-A0CU-01Z-00-DX1
77,484
53,882
402
400
6
false
End of preview. Expand in Data Studio

Pan-Cancer-Nuclei-Seg — Manual Subset

Gold-standard manual nucleus segmentation patches from the TCIA Pan-Cancer-Nuclei-Seg resource (Hou et al., Scientific Data 2020): 1,356 manually segmented 256×256 H&E patches with per-nucleus instance masks, spanning 14 TCGA cancer types.

Scope — please read. This repository contains only the manual subset, not the full Pan-Cancer-Nuclei-Seg resource. The headline "~5 billion nuclei" component (5,060 whole-slide images, ~666 GB of CSV polygon vertices) is algorithm-generated (a U-Net pipeline, not manual ground truth), is distributed without source images (the H&E WSIs live in TCGA/GDC), and is not hosted here. The 1,356 manual patches are the only part that is both gold-standard and image-paired.

  • Modality: Histopathology — H&E brightfield, 256×256 patches at 40× (~0.25 µm/px)
  • Target: nucleus segmentation (single foreground class)
  • Ground truth: manually corrected Mask R-CNN masks agreed by annotators A, B and C collectively (per the dataset readme)
  • License: CC BY 3.0
  • Source: TCIA Pan-Cancer-Nuclei-Seg · Stony Brook BMI Box cnn-nuclear-segmentations-2019

Cancer types (14)

BLCA, BRCA, CESC, COAD, GBM, LUAD, LUSC, PAAD, PRAD, READ, SKCM, STAD, UCEC, UVM (~91–100 patches each, balanced).

Columns

Column Type Notes
patch_id int32 Patch-ID from the dataset readme (non-contiguous, 1..1365 with gaps)
image Image (RGB) 256×256 H&E patch ({id}_crop.png), original PNG bytes
instance_mask Image (16-bit, mode I;16) Consensus manual mask ({id}_labeled_mask_corrected.png); 0 = background, 1..N = per-nucleus instance IDs. Original PNG bytes (lossless)
cancer_type string One of the 14 TCGA codes (lowercase)
wsi_id string TCGA slide barcode the patch was cropped from (e.g. TCGA-2F-A9KR-01Z-00-DX1) — use as the cross-dataset dedup key
x, y int32 Top-left coordinate of the crop in the source WSI
size_original int32 Crop side length in source-WSI pixels before resize
size_in_40x int32 Crop side length at 40× (400)
num_instances int32 Number of nuclei in the patch (= instance_mask.max())
has_multirater bool True for the 27 patches that also have per-annotator masks at the source

For binary nucleus-vs-background segmentation, treat instance_mask > 0 as foreground. Read the mask via numpy (np.array(mask)) — passing a 16-bit I;16 PNG through PIL.Image.convert("L") divides values by 256 and erases most instance IDs.

Empty masks: 32 of the 1,356 patches contain no annotated nuclei (num_instances == 0, background/stroma-only tissue) — these are faithful to the source, not a packaging error. Filter on num_instances > 0 if your task requires at least one nucleus (instance counts range 0–141, mean ≈ 29).

Provenance, naming and cross-dataset overlap

  • Provenance: official, author-hosted (Saltz/Kurc, Stony Brook BMI). 1,356 patches matches the paper.
  • Faithful naming: this is the manual subset only (see scope note above).
  • Ground-truth tier: the bulk Pan-Cancer-Nuclei-Seg masks are algorithm-generated; only these 1,356 patches are manual. 27 patches carry additional per-annotator masks at the source.
  • Overlap (leakage hazard): every patch is TCGA-derived, so it shares source slides with other TCGA histopathology sets — notably PanNuke, MoNuSeg, MoNuSAC, NuCLS and TIGER (on BRCA/PRAD/BLCA/LUAD). Deduplicate against those by matching the wsi_id TCGA barcode before any joint benchmark.

Citation

@article{hou2020pancancernucleiseg,
  title   = {Dataset of segmented nuclei in hematoxylin and eosin stained
             histopathology images of ten cancer types},
  author  = {Hou, Le and Gupta, Rajarsi and Van Arnam, John S. and Zhang, Yuwei
             and Sivalenka, Kaustubh and Samaras, Dimitris and Kurc, Tahsin M.
             and Saltz, Joel H.},
  journal = {Scientific Data},
  volume  = {7},
  number  = {1},
  pages   = {185},
  year    = {2020},
  doi     = {10.1038/s41597-020-0528-1}
}

TCIA data citation: Hou, L. et al. (2019). Dataset of Segmented Nuclei in Hematoxylin and Eosin Stained Histopathology Images of Ten Cancer Types (Pan-Cancer-Nuclei-Seg). The Cancer Imaging Archive. doi:10.7937/TCIA.2019.4A4DKP9U

Downloads last month
24