ISLES'24 : a real-world longitudinal multimodal stroke dataset

This multi-center dataset consists of 149 acute ischemic stroke cases, representing the training set of the ISLES'24 challenge. All data are provided in NIfTI format (.nii.gz) and organized according to the BIDS standard. For each case, the following data are included: - Admission imaging: non-contrast CT (NCCT), CT angiography (CTA), 4D CT perfusion (CTP) time series, and perfusion maps (Tmax, CBF, CBV, MTT). - Follow-up imaging: post-treatment MRI (DWI and ADC). - Clinical data: demographics, patient history, admission NIHSS, 3‑month functional outcome (mRS), etc. - Annotations: binary infarct masks derived from follow-up MRI (lesion-msk.nii.gz), large vessel occlusion binary masks derived from CTA (lvo-msk.nii.gz), and the multi-labeled Circle of Willis anatomy generated with an automatic algorithm over CTA (cow-msk.nii.gz) This dataset combines multimodal imaging, longitudinal follow-up, and structured clinical variables to support benchmarking of stroke infarct prediction methods.