Table 1.
Summary of the datasets used in this study.
| Dataset | Composition | Purpose |
|---|---|---|
| BRIGHT | 2 169 355 Patches from 50 WSIs 787 168 - Epithelium, 863 989 - Stroma 98 293 - Lymphocytes, 245 525 Adipose 127 393 - Artifacts, 46 987 - Miscellaneous |
Training and validation of HistoROI |
| CRC-100k | 100 000 patches of colon cancer 10 407 - ADI, 8763 - NORM, 10 566 - BACK, 11 557 - LYM, 14 317 - TUM, 13 536 - MUS, 8896 - MUC, 10 446 - STR, 11 512 - DEB |
Validation of HistoROI on external dataset |
| TCGA-4Org | 93 multi-organ WSIs annotated with annotated tissue region | Application of HistoROI for QC |
| CAMELYON16 | 399 WSIs - 270/129 in train/test set Train - 70/100 positive/negative WSIs Test - 49/80 positive/negative WSIs |
Application of HistoROI as WSI pre-processing |
| TCGA-Lung | 1034 WSIs - 634/404 in train/test set Train - 316/318 adeno/squamous WSIs Test - 211/193 adeno/squamous WSIs |
Application of HistoROI as WSI pre-processing |