Frequency-Aware Ensemble Learning for BraTS 2025 Pediatric Brain Tumor Segmentation
This work addresses the critical problem of accurate segmentation for rare and heterogeneous pediatric brain tumors, which is incremental as it builds on existing models with specific enhancements.
The paper tackled pediatric brain tumor segmentation for the BraTS 2025 challenge by proposing an ensemble method integrating nnU-Net, Swin UNETR, and HFF-Net, achieving first place with Dice scores up to 92.6% on unseen test data.
Pediatric brain tumor segmentation presents unique challenges due to the rarity and heterogeneity of these malignancies, yet remains critical for clinical diagnosis and treatment planning. We propose an ensemble approach integrating nnU-Net, Swin UNETR, and HFF-Net for the BraTS-PED 2025 challenge. Our method incorporates three key extensions: adjustable initialization scales for optimal nnU-Net complexity control, transfer learning from BraTS 2021 pre-trained models to enhance Swin UNETR's generalization on pediatric dataset, and frequency domain decomposition for HFF-Net to separate low-frequency tissue contours from high-frequency texture details. Our final ensemble framework combines nnU-Net ($γ=0.7$), fine-tuned Swin UNETR, and HFF-Net, achieving Dice scores of 62.7% (CC), 83.2% (ED), 72.9% (ET), 85.7% (NET), 91.8% (TC), and 92.6% (WT) on the unseen test dataset, respectively. Our proposed method achieves first place (rank 1st) in the BraTS 2025 Pediatric Brain Tumor Segmentation Challenge.