DynFS-MoE: Dynamic Functional-Structural Mixture-of-Experts for Post-Traumatic Epilepsy Diagnosis
Researchers have developed a new dynamic multimodal Mixture-of-Experts (MoE) framework called DynFS-MoE to improve the early diagnosis of post-traumatic epilepsy (PTE). This framework integrates functional and structural MRI data by using time-aware encoding and class-conditioned expert routing. Experimental results show that DynFS-MoE outperforms static fusion methods and provides interpretable insights into region-of-interest interactions, offering a more effective approach for PTE diagnosis and risk stratification. AI
IMPACT This new framework could lead to earlier and more accurate diagnoses of post-traumatic epilepsy, improving patient outcomes.