Ischemic Stroke Lesion Segmentation Challenge (ISLES'26)¶

Multi-phase infarct segmentation in ~2,000 T1w MRI¶

🔬 Background¶

The ISLES 2026 edition establishes a robust benchmark for generalized stroke infarct segmentation on T1-weighted MRI scans. This initiative directly addresses the generalization gaps identified in the previous ISLES’22-ATLAS (a prior attempt to this task we organized). Participants will develop methodologies on one of the largest-ever annotated stroke infarct datasets.

Key Highlights¶

• Massive Scale: ~2,000 annotated scans.
• Diversity: Data sourced from 60+ international centers.
• Comprehensive Coverage: Includes Acute, Sub-acute, and Chronic disease stages.
• Expansion: Provides a 40% increase in data diversity and a greater than 2-fold expansion in training data.

🎯 The Task¶

Participants are challenged to develop automated segmentation algorithms that accurately delineate ischemic stroke lesions in T1-weighted MRI, covering acute, sub-acute, and chronic lesions.

Inputs¶

• Native space T1w MRI
• Metadata (.json): Includes stroke lesion covariates such as DAYS_POST_STROKE, CHRONICITY (1 if < 180 days), and CENTER code.

Outputs¶

• Binary infarct segmentation mask.

📅 Schedule¶

📝 Registration¶

[!IMPORTANT] Verified accounts are mandatory to be accepted to the challenge. Please ensure your Grand-Challenge profile is complete and verified before requesting to join.