The ISLES'24 Dataset: A Multimodal Stroke Imaging Dataset with Hyperacute CT, Acute Postinterventional MRI, and 3-month Clinical Outcomes.
Authors
Affiliations (16)
Affiliations (16)
- Department of Diagnostic and Interventional Neuroradiology, School of Medicine and Health, TUM Klinikum Rechts der Isar, Technical University of Munich, Ismaninger Strasse 22, Munich 81675, Germany.
- Department of Quantitative Biomedicine, University of Zurich, Zurich, Switzerland.
- Department of Neurology, University Hospital of Zurich and University of Zurich, Zurich, Switzerland.
- Center for Computational Health, Zurich University of Applied Sciences, Zurich, Switzerland.
- icometrix, Leuven, Belgium.
- Department of ??????, University of Girona, Girona, Spain.
- Support Center of Advanced Neuroimaging (SCAN), University Institute of Diagnostic and Interventional Neuroradiology, Inselspital, Bern, Switzerland.
- University Institute of Diagnostic and Interventional Neuroradiology, University Hospital Bern, Inselspital, University of Bern, Bern, Switzerland.
- ARTORG Center for Biomedical Research, University of Bern, Bern, Switzerland.
- Department of Radiation Oncology, University Hospital Bern, University of Bern, Switzerland.
- Department of Biomedical Engineering, Eindhoven University of Technology, Eindhoven, the Netherlands.
- Chair for AI in Healthcare and Medicine, Technical University of Munich (TUM) and TUM University Hospital, Munich, Germany.
- Department of Computing, Imperial College London, London, United Kingdom.
- TranslaTUM - Central Institute for Translational Cancer Research, Technical University of Munich, Munich, Germany.
- AI for Image-Guided Diagnosis and Therapy, School of Medicine and Health, Technical University of Munich, Munich, Germany.
- Department of Neurology, University Hospital Zurich, Zurich, Switzerland.
Abstract
Stroke remains a major global health burden (1,2), although outcomes have improved substantially through imaging-guided therapy and endovascular reperfusion (3,4). While CT and MRI are standard for estimating infarct core and penumbra (5), variability in threshold-based deconvolution of perfusion imaging (6) can lead to inconsistent lesion size estimates (7). Accurate modeling of infarct growth is therefore essential for optimizing transfer and treatment decisions (8). Advances in artificial intelligence (AI) have improved automated lesion detection, yet clinical translation requires large, well-annotated datasets. While recent large-scale cohorts including the Ischemic Stroke Lesion Segmentation Challenge (ISLES)'22 (<i>n</i> = 400) (9), Liew et al (<i>n</i> = 1271) (10), Liu et al (<i>n</i> = 2888) (11), and Absher et al (<i>n</i> = 1715) datasets (12) have expanded available imaging data, datasets pairing acute CT with follow-up MRI (13) remain limited. We address this gap by providing a publicly available dataset that combines hyperacute CT (< 24 h post onset) with acute postinterventional MRI (2-9 days after successful reperfusion; modified Treatment in Cerebral Ischemia 2c or 3) and structured clinical follow-up through 3 months. This combination enables analysis of infarct evolution and supports AI model development for postinterventional stroke care. © RSNA, 2026.