How We Won the ISLES'24 Challenge by Preprocessing

TL;DR

This paper presents a preprocessing pipeline combined with a residual nnU-Net architecture that effectively segments stroke lesions from CT scans, winning the ISLES'24 challenge by predicting lesion progression with limited input data.

Contribution

The authors introduce a novel preprocessing approach including deep-learning skull stripping and intensity windowing that enhances segmentation accuracy in a challenging CT-based stroke lesion dataset.

Findings

Achieved a mean test Dice score of 28.5

Preprocessing significantly improved segmentation performance

Winning solution for the ISLES'24 challenge

Abstract

Stroke is among the top three causes of death worldwide, and accurate identification of stroke lesion boundaries is critical for diagnosis and treatment. Supervised deep learning methods have emerged as the leading solution for stroke lesion segmentation but require large, diverse, and annotated datasets. The ISLES'24 challenge addresses this need by providing longitudinal stroke imaging data, including CT scans taken on arrival to the hospital and follow-up MRI taken 2-9 days from initial arrival, with annotations derived from follow-up MRI. Importantly, models submitted to the ISLES'24 challenge are evaluated using only CT inputs, requiring prediction of lesion progression that may not be visible in CT scans for segmentation. Our winning solution shows that a carefully designed preprocessing pipeline including deep-learning-based skull stripping and custom intensity windowing is beneficial for accurate segmentation. Combined with a standard large residual nnU-Net architecture for segmentation, this approach achieves a mean test Dice of 28.5 with a standard deviation of 21.27.