What will appear here
The catalogue is still empty: we add a dataset only after checking its origin and license. Each one will get a Data Card covering:
- where the data comes from and on what terms it can be used;
- modality and task;
- whether it may be redistributed;
- how it was de-identified;
- the patient-level split and data version.
Downloading is not redistributing
Where a license forbids redistribution, we publish download scripts, checksums, converters and fixed splits instead of the data. A dataset being publicly available does not give us the right to share a copy.
Read more in ethics and licenses. You can propose a dataset through a repository issue: cite public sources only and never attach patient files.