ISLES 2022: A Multi-Center Magnetic Resonance Imaging Stroke Lesion Segmentation Dataset

Creators

Moritz Roman Hernandez Petzsche¹
Ezequiel de la Rosa²
Roland Wiest³
Uta Hanning⁴
Benedikt Wiestler¹
Jan S. Kirschke¹

Description

This multi-center dataset includes 250 expert-annotated stroke MRI cases and serves as the official training dataset for the Ischemic Stroke Lesion Segmentation Challenge (ISLES'22).

For each case, expert annotations of stroke lesions are provided along with the following imaging sequences:

FLAIR – Fluid Attenuated Inversion Recovery
DWI – Diffusion Weighted Imaging (b=1000)
ADC – Apparent Diffusion Coefficient map

All imaging data and annotations are provided in NIfTI format (nifti.nimh.nih.gov) and follow the BIDS convention. Images are in native space (not registered) and have been skull-stripped to de-identify patients.

Image Acquisition

Scans were performed using the following MRI devices:

3T Philips MRI scanners: Achieva, Ingenia
3T Siemens Verio
1.5T Siemens MAGNETOM scanners: Avanto, Aera

All images were acquired during routine clinical care for stroke patients at three stroke centers and collected retrospectively as part of various clinical studies.

Where available, scanner metadata from DICOM headers is provided in .json format.

More Information

For full dataset details, refer to the ISLES'22 preprint.
Challenge details available on the official ISLES'22 website and the Grand Challenge page.

Citation

Please cite the following publication when using this dataset:

Hernandez Petzsche, M.R., de la Rosa, E., Hanning, U. et al. ISLES 2022: A multi-center magnetic resonance imaging stroke lesion segmentation dataset. Sci Data 9, 762 (2022).
https://doi.org/10.1038/s41597-022-01875-5