Stroke lesion segmentation improves with size adaptive decision layer
SADL: A Size-Adaptive Decision Layer for Chronic Stroke Lesion Segmentation in ISLES 2026
Computer Vision and Pattern Recognition
Summary
Stroke lesions in the brain are hard to identify accurately on MRI scans because they can look similar to healthy tissue and vary greatly in size. The authors created a system that combines multiple neural networks and uses a special decision layer, called SADL, which adjusts detection thresholds based on lesion size. This method helps find small stroke lesions that might be missed otherwise. Their approach improves the accuracy of identifying stroke damage in brain scans without using any data from the test set during tuning.
What this means in practice
- •For medical imaging teams: Enhance chronic stroke lesion detection on MRI by using an adaptive decision layer that improves accuracy especially for small lesions.
- •For clinical software developers: Incorporate size-adaptive thresholding in automated stroke lesion segmentation tools to reduce missed small lesions and support better clinical decisions.
Authors
Marek Wodzinski, Gniewosz Drwiega, Wojciech Szymanski
Abstract
Chronic ischemic stroke lesions are difficult to segment on T1-weighted MRI. Their appearance overlaps with healthy tissue, and their volumes span four orders of magnitude, so voxel-level and lesion level metrics often disagree. We build an eleven-member nnU-Net ensemble that covers five design axes: a baseline dice + top-K loss, heavy augmentation, per-lesion volume loss re-weighting, stochastic weight averaging, and a sampling-and-architecture block that pairs cohort-balanced sampling with a residual-encoder backbone. Weighted soft-map fusion of the members feeds a tuned decision layer, SADL, that estimates lesion volume, applies an adaptive per-lesion-size threshold, prunes components by voxel count and peak probability, and adds a never-empty rescue. All fusion weights and per-bucket parameters are fitted on out-of-fold (OOF) predictions, so tuning never touches the held-out test set. On a site-disjoint 91-case holdout, SADL reaches Dice 0.7322, lesion-F1 0.7461 and PR-AUC 0.8430 when scored online with the challenge's released evaluation code. A paired non-parametric bootstrap against a reference single ensemble member confirms the lesion-F1 gain at p=0.024, and the gain is largest in the small-lesion regime, where the rescue step recovers cases that a fixed threshold would silently discard.