ISLES 2016 (Ischemic Stroke Lesion Segmentation Challenge 2016)

ISLES 2016 – Ischemic Stroke Lesion Segmentation Challenge 2016 Brain-MRI ischemic-stroke dataset from the MICCAI 2016 ISLES challenge. This is a lesion-OUTCOME-PREDICTION task, not conventional visible-lesion segmentation: the inputs are acute perfusion-derived MR maps, while the ground-truth mask is the final infarct traced on a ~90-day follow-up scan and mapped back into the acute imaging space. Read before benchmarking. A model is asked to predict future tissue fate from… See the full description on the dataset page:

Source: Hugging Face Hub (MedOtter/isles2016). Metadata imported from the dataset’s Hub tags.