M\textsuperscript{4}Fuse: Lightweight State-Space MoE with a Cross-Scale Gating Bridge for Brain Tumor Segmentation

2026-05-04

AI Summary

This work addresses the high computational cost and limited robustness in 3D brain tumor segmentation arising from encoder–decoder imbalance and large input volumes. To this end, the authors propose a lightweight collaborative architecture that balances encoder and decoder capacities, incorporates a grouped state space mixer to model long-range contextual dependencies, and introduces a cross-scale, two-stage gated bridge to denoise and align skip connections. Furthermore, a sample-level mixture-of-experts (MoE) mechanism is integrated to enhance multi-center generalization. Evaluated on BraTS2019 and BraTS2021 at a resolution of 64×128×128, the method achieves a 62.63% reduction in parameter count while improving the average Dice score by 0.09, significantly outperforming existing lightweight approaches.

Abstract

Encoder-decoder imbalance and the reliance on large input volumes make many 3D brain tumor segmentation models both compute-heavy and brittle. We present M\textsuperscript{4}Fuse, a lightweight network that prioritizes discriminative brain tumor cues over exhaustive appearance reconstruction. Our method balances encoder and decoder capacity and replaces depth expansion with a synergistic design: it propagates long-range context with linear complexity via a grouped state space mixer, denoises and aligns skip features using a cross-scale dual-stage gating bridge, and absorbs cross-site acquisition shifts with a sample-level mixture-of-experts. On the BraTS2019 and BraTS2021 benchmarks, M\textsuperscript{4}Fuse outperforms other lightweight excellent methods in both parameter count and performance. Even at a challenging input resolution of \(64\times128\times128\) (half that of existing excellent models), M\textsuperscript{4}Fuse reduces parameters by 62.63\% and improves average performance by 0.09\%. Ablations of key components validate the method's exceptional parameter-to-accuracy efficiency and robustness across diverse data centers.

Problem

• brain tumor segmentation
• encoder-decoder imbalance
• compute-heavy models
• input volume dependency
• model brittleness

Innovation

• state-space model
• mixture-of-experts
• cross-scale gating
• lightweight segmentation
• brain tumor segmentation