README.md

metadata

license: cc-by-nc-nd-4.0
tags:
  - medical
  - pathology
  - segmentation

PathSeg Dataset

This repository provides the official release of the PathSeg dataset used in the paper:

“Segment Anything in Pathology Images with Natural Language.”

PathSeg is released to support research on pathology image understanding and segmentation with natural language prompts.

Only the subsets for which explicit distribution permissions were obtained from the respective data owners are included in this release.

For convenience, we list below the original source datasets from which PathSeg is derived. Please refer to the corresponding URLs for the original downloads and license terms.

Dataset	URL
BCSS	https://github.com/PathologyDataScience/BCSS
CoCaHis	https://cocahis.irb.hr/
CoNIC2022	https://conic-challenge.grand-challenge.org/
CRAG	https://opendatalab.com/OpenDataLab/CRAG/
DigestPath2019	https://digestpath2019.grand-challenge.org/
GlaS	https://datasetninja.com/gland-segmentation
Janowczyk	https://andrewjanowczyk.com/use-case-1-nuclei-segmentation/
MIDOG	https://midog.deepmicroscopy.org/download-dataset/
MoNuSAC	https://monusac-2020.grand-challenge.org/Data/
NuCLS	https://sites.google.com/view/nucls/home
PanNuke	https://www.kaggle.com/datasets/andrewmvd/cancer-inst-segmentation-and-classification
RINGS	https://data.mendeley.com/datasets/h8bdwrtnr5/1
SegPath	https://dakomura.github.io/SegPath/
SICAPv2	https://www.kaggle.com/datasets/shridharspol/sicapv2
TNBC	https://peterjacknaylor.github.io/data/
WSSS4LUAD	https://wsss4luad.grand-challenge.org/
CoNSeP	https://opendatalab.com/OpenDataLab/CoNSeP
CPM15	https://drive.google.com/drive/folders/11ko-GcDsPpA9GBHuCtl_jNzWQl6qY_-I
CPM17	https://drive.google.com/drive/folders/1sJ4nmkif6j4s2FOGj8j6i_Ye7z9w0TfA
Kumar	https://drive.google.com/drive/folders/1bI3RyshWej9c4YoRW-_q7lh7FOFDFUrJ
Lizard	https://www.kaggle.com/datasets/aadimator/lizard-dataset

Citation

Contact

For questions, permissions, or issues related to the dataset, please contact: zchenhi@connect.ust.hk.