openmedlab/awesome-medical-dataset

Purpose and Entry Point

This is a static “awesome list” repository cataloging public medical datasets, benchmarks, related methods, and external tooling. The primary reader entry point is README.md, which provides the navigable taxonomy, dataset summaries, and links to detailed records.

Catalog Index

README.md is the curated index and presentation layer. It groups datasets by imaging anatomy/modality, image-text, and text datasets, then links each dataset name to a dedicated Markdown record. It also contains outbound links to benchmarks, SOTA methods, dataset platforms, and related repositories.

Dataset Record Store

Each dataset is represented by an independently maintained Markdown document under resources/. These files are the repository’s durable content units and are referenced from the root catalog:

• resources/CT-ORG.md

• resources/TotalSegmentator.md

• resources/MedQA.md

The boundary is intentionally documentation-only: records describe and link to datasets rather than containing the datasets themselves.

Domain Taxonomy

The root index organizes imaging datasets into clinical/anatomical collections such as whole body, head and neck, chest, abdomen, heart, bones, endoscopy, retina, skin, and microscopic imaging. It separately models multimodal and text-only resources. Representative records spanning those divisions include:

• resources/BraTS2021.md

• resources/Cholec80.md

• resources/PMC-VQA.md

This taxonomy is implemented through Markdown headings, tables, anchors, and relative links rather than code or a data schema.

Supplemental Assets and Exceptions

resources/appendix/ contains supplementary materials, primarily image/media assets associated with selected dataset entries, plus an additional Markdown record:

• resources/appendix/3D-IRCADB_0.avif

• resources/appendix/SegPC21_0.avis

• resources/appendix/BONBID-HIE.md

These assets support documentation presentation; the main catalog does not contain application logic for processing them.

External-Service Boundary

The repository delegates dataset hosting, downloads, challenges, and documentation to external providers. Linked services include Grand Challenge, Kaggle, TCIA, Synapse, CodaLab/Codabench, GitHub, Hugging Face, Zenodo, and Tianchi. Individual resource files and README.md contain those outbound URLs; no credentials, APIs, deployment infrastructure, or executable runtime are present.

Maintenance Model

Content changes are editorial: add or update a resources/*.md dataset record, then add or revise its categorized row/link in README.md; optional associated media belongs under resources/appendix/. There are no visible source modules, build pipeline, tests, package manifests, or server/client entry points in the repository tree.