π€ AI Summary
This study addresses the significant performance degradation of post-operative glioma segmentation under cross-institutional clinical protocols due to domain shift. To mitigate this issue, the authors propose brain-mask percentile normalization combined with voxel-level contrastive learning to stabilize training dynamics. Furthermore, they introduce a Subspace-Aware Class Attention (SACA) module to recalibrate bottleneck features and enhance sensitivity to contrast-enhancing tumor regions. Integrated into the nnU-Net framework, the proposed method achieves a Dice coefficient of 0.94 for whole lesion segmentation on the MU-GLIOMA-POST dataset. The SACA-augmented variant reduces boundary error to an HD95 of 2.92 mm and yields a relative 9.1% improvement in sensitivity to enhancing tumor regions.
π Abstract
Tracking residual tumor after surgery is essential for catching recurrence early, but automating post-operative glioma segmentation remains a difficult task. Although transformer-based architectures, such as SwinUNETR, achieved impressive results, few studies test how well they generalize across clinical protocols. In this paper, we conduct an ablation study on the MU-GLIOMA-POST and UCSF-ALPTDG datasets and show that the standard Generalized Dice Loss (GDL) is unstable under domain shift: the Whole Lesion (WL) Dice drops from 0.88 on the internal validation set to 0.73 on the external UCSF test set. To address this, we pair brain-masked percentile normalization with voxel-level contrastive learning. We also propose a Subspace-Aware Class Attention (SACA) module that re-calibrates the bottleneck features and raises Enhancing Tumor (ET) sensitivity by 8% (9.1% relative improvement) on internal validation. Ensembling these refinements with nnU-Net brings every stable configuration to a WL Dice of 0.94, and the SACA variant ensemble achieves the best boundary error (HD95) of 2.92 mm on MU-GLIOMA-POST.