PALM (PAthoLogic Myopia, iChallenge 2019): 1,200 fundus photographs with optic disc, atrophy and detachment masks

Provider
Huihui Fang, Fei Li, Junde Wu, Huazhu Fu et al. (iChallenge PALM); mirrored by MedOtter
License
CC0-1.0 (source)
Size
12 files
Format
Parquet (in sampled files)
Stored at
Hugging Face (https://huggingface.co/datasets/MedOtter/iChallenge-PALM19)
Access
Free and public. No account with the data provider is needed.
DOI
10.6084/m9.figshare.21299148.v1
Category
Health & Biomedical
Source
https://doi.org/10.6084/m9.figshare.21299148.v1

About this dataset

PALM (PAthoLogic Myopia) is the iChallenge 2019 dataset from Zhongshan Ophthalmic Center, Sun Yat-sen University: 1,200 color fundus photographs of the left eyes of 720 myopia clinic patients, taken on a Zeiss Visucam 500 (2124x2056) or Canon CR-2 (1444x1444). Each image is labeled pathologic myopia or not (637 versus 563), and pixel masks are given for the optic disc (1,144 images), patchy retinal atrophy (870) and retinal detachment (24). Splits are patient-disjoint with 400 images each in train, validation and test. Images keep their original JPEG bytes and masks are binary PNG, stored as Parquet under data/, about 1.7 GB (measured 2026-11-21); read with Hugging Face datasets or pandas plus Pillow. Use for pathologic myopia classification, optic disc segmentation, retinal atrophy lesion segmentation and fovea localization. Mirrored by MedOtter from the authors' figshare deposit, released under CC0 1.0 with DOI 10.6084/m9.figshare.21299148.v1; cite Fang et al. and the figshare deposit.

How to load this dataset in Python

from datasets import load_dataset
ds = load_dataset("MedOtter/iChallenge-PALM19")
print(ds)

Data structure

File types in sampled files: Parquet

Files

Example file: data/test-00000-of-00003.parquet

Analyze this dataset without downloading it

On Scigantic, PALM (PAthoLogic Myopia, iChallenge 2019): 1,200 fundus photographs with optic disc, atrophy and detachment masks opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.

This page is also available as Markdown and as JSON.

Related Health & Biomedical datasets