2026-10-06T06:28:20Z https://api.figshare.com/v2/oai

oai:figshare.com:article/32646378 2026-09-27T19:39:04Z category_24817 category_28864 item_type_3 month_year_09_2026

A multi-disease fundus image segmentation dataset with pixel-level structural annotations Zhiling Li (24173421) Ophthalmology Artificial intelligence not elsewhere classified Color fundus photography Fundus image segmentation Retinal image analysis <p dir="ltr">This dataset comprises 7,720 color fundus photography (CFP) image IDs with pixel-level structural annotations, compiled from 11 publicly available datasets and one clinical dataset provided by The Second Affiliated Hospital, Zhejiang University School of Medicine. ScienceDB directly distributes 5,772 CFP images and structural masks for all 7,720 image IDs. The remaining 1,948 images must be obtained from their official source datasets and matched to the released masks using the provided mapping file. Each image ID has three annotation types: binary vessel masks, artery–vein classification masks, and optic disc/cup masks.</p><p dir="ltr">The dataset is divided into 26 independently downloadable disease-specific archives. Each archive contains the three types of masks for the corresponding image IDs and, where redistribution is permitted, the associated fundus images. Files follow a unified naming convention. The record also provides `hard_cases.xlsx`, `source_datasets.xlsx`, and `source_image_mapping.xlsx`. In `hard_cases.xlsx`, task-specific flags identify image–task pairs for which severe pathology prevented reliable manual annotation; their uncorrected AI masks are retained only for reference.</p><p dir="ltr">This dataset supports retinal vessel segmentation, artery–vein classification, optic disc/cup segmentation, multi-task learning, multi-disease modeling, and ophthalmic AI research. Please cite the associated Data Descriptor and the relevant original source datasets when using this resource.</p> 2026-09-27T19:39:04Z Dataset Dataset 10.6084/m9.figshare.32646378.v1 https://figshare.com/articles/dataset/A_multi-disease_fundus_image_segmentation_dataset_with_pixel-level_structural_annotations/32646378 CC BY 4.0