datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
cross-channel-nuclei-from-cellmask
Cross-channel nuclei-from-cellmask dataset
3029 paired fields. images/ is the
cell-mask input channel, masks/ the matching instance-labelled nuclei ground truth (same stem =
same field). split_by_well.csv gives the train/test assignment; the split is by well, so no
well contributes to both sides - field-level splitting would leak.
Prepared with spaCR
(PyPI | conda-forge).
Used to train einarolafsson/cross-channel-nuclei-from-cellmask-cpsam.
Pan-Cancer-Nuclei-Seg
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… See the full description on the dataset page: https://huggingface.co/datasets/MedOtter/Pan-Cancer-Nuclei-Seg.NucleiData