Hyper-Connections for Adaptive Multi-Modal MRI Brain Tumor Segmentation
This work addresses the problem of improving segmentation accuracy for brain tumors in medical imaging, offering an incremental enhancement to existing methods.
The study tackled brain tumor segmentation from multi-modal MRI by introducing Hyper-Connections as a drop-in replacement for fixed residual connections, achieving up to a +1.03 percent mean Dice gain across five 3D architectures on the BraTS 2021 dataset.
We present the first study of Hyper-Connections (HC) for volumetric multi-modal brain tumor segmentation, integrating them as a drop-in replacement for fixed residual connections across five architectures: nnU-Net, SwinUNETR, VT-UNet, U-Net, and U-Netpp. Dynamic HC consistently improves all 3D models on the BraTS 2021 dataset, yielding up to +1.03 percent mean Dice gain with negligible parameter overhead. Gains are most pronounced in the Enhancing Tumor sub-region, reflecting improved fine-grained boundary delineation. Modality ablation further reveals that HC-equipped models develop sharper sensitivity toward clinically dominant sequences, specifically T1ce for Tumor Core and Enhancing Tumor, and FLAIR for Whole Tumor, a behavior absent in fixed-connection baselines and consistent across all architectures. In 2D settings, improvements are smaller and configuration-sensitive, suggesting that volumetric spatial context amplifies the benefit of adaptive aggregation. These results establish HC as a simple, efficient, and broadly applicable mechanism for multi-modal feature fusion in medical image segmentation.