GMENet: Generative Mixture of Experts Network for Multi-Center Glioma Diagnosis with Incomplete Imaging Sequences

2026-05-21

AI Summary

This study addresses the challenge posed by missing MRI sequences in clinical settings due to protocol heterogeneity across centers, which hinders the effective use of incomplete multi-center data by existing models. To overcome this limitation, the authors propose GMENet, a novel framework that synthesizes missing sequence features via a cross-attention gated generation module and fuses original with generated features through a dynamic weighted mixture-of-experts (MoE) mechanism. This approach preserves semantic consistency while enhancing robustness to both missing data and cross-center distribution shifts. The method incorporates cycle-consistency loss and a confidence-aware fusion strategy, and is validated on a multi-center glioma dataset comprising 1,241 cases. Results demonstrate a 97% increase in usable training data and superior performance compared to state-of-the-art methods trained on complete data alone.

Abstract

Contemporary glioma diagnosis integrates molecular features with histopathology to guide clinical decision-making. However, in clinical settings, divergent imaging protocols result in incomplete MRI sequences, leading to two primary challenges: forcing existing frameworks to discard a large portion of clinical data during training and consequently limiting their clinical applicability. To address these limitations, we propose GMENet, a Generative Mixture of Experts Network for multi-center glioma diagnosis with incomplete imaging sequences. Firstly, we design a Cross-attention-based Gated Generation Module that synthesizes missing sequence features from available sequences via cross-attention and dynamic gating mechanisms, incorporating a cycle-consistency loss to preserve semantic integrity. Secondly, we introduce a Dynamically Weighted Experts Fusion Module that performs mixture-of-experts interaction and confidence-aware fusion over original and synthesized dual-sequence features for multi-task prediction. We evaluate GMENet on a multi-center cohort of 1,241 subjects from four in-house datasets and two public repositories. Experiments show that GMENet expands clinically usable training data by 97\%, relative to complete-sequence-only data. Furthermore, it consistently outperforms state-of-the-art methods trained on complete data, demonstrating improved robustness under cross-center distribution shifts.

Problem

• incomplete MRI sequences
• multi-center glioma diagnosis
• clinical data usability
• imaging protocol divergence

Innovation

• Generative Mixture of Experts
• Cross-attention Generation
• Incomplete MRI Sequences
• Multi-center Glioma Diagnosis
• Dynamic Expert Fusion