Adaptive Post-Processing Drives Instance-Level Detection in Stroke Lesion Segmentation
2026-08-17 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionArtificial Intelligence
AI summaryⓘ
The authors focused on improving how well medical image tools detect individual small lesions, not just how much their segmentations overlap with true lesions. They found that adjusting the way results are filtered after the model runs (post-processing) matters more for detecting small lesions than changing the model itself. Their new method, VCAP, tweaks detection thresholds based on lesion sizes and improves lesion detection scores noticeably, while a special model architecture mostly improved detection but not overlap scores. Combining their methods led to better performance than traditional approaches using just one model without post-processing.
lesion detectionmedical image segmentationpost-processingvoxel overlapsmall lesionsDice scoreLesion-F1adaptive thresholdingensemble modelISLES challenge
Authors
Qinghui Liu, Jon André Ottesen, Atle Bjørnerud, Kyrre Eeg Emblem
Abstract
Instance-level lesion detection has been an increasingly larger focal point in medical image segmentation besides the more standard voxel-level overlap. Still, most pipelines are trained and post-processed for voxel overlap alone. In particular, the mismatch is most pronounced for small lesions, where a near-miss prediction---substantial overlap that falls just short of the instance-matching threshold---scores the same as a complete miss. In our ISLES'26 submission, we found that closing this gap mattered far more in post-processing than in architecture design. Our Volume-Conditioned Adaptive Post-Processing (VCAP) scheme adjusts component-size thresholds to each case's predicted lesion burden, improving Lesion-F1 by 0.032 (unbiased cross-fold estimate)---approximately 6 times larger than any architectural change we tested. A resolution-aware attention architecture (Viola2Plus), designed for small-lesion segmentation, shows why the distinction matters: it left small-lesion Dice unchanged but raised small-lesion detection rate by 3.7\%, a real effect voxel-overlap metrics alone would have missed. Under 5-fold cross-validation on the 1,453-case training set, our post-processed two-architecture ensemble achieves Dice 0.651 and Lesion-F1 0.614, versus 0.644 and 0.573 for the unprocessed single-model baseline.