SCNN: Spike Coupling Neural Network for Multimodal Brain Network Analysis

Abstract

Structure-function coupling (SC-FC coupling) refers to the correlation between the physical layout of structural connectivity (SC) in the brain and the activity patterns of functional connectivity (FC). However, most existing SC-FC coupling analysis methods lack system-level integration and overlook the neurobiological mechanisms between SC and FC. Spike Neural Network (SNN) is a computational model inspired by the biological nervous system, in which the dense interconnection of spiking neurons and synaptic weights can effectively simulate the interaction between SC and FC in the brain. In this paper, we propose a novel spike coupling neural network (SCNN) for multimodal brain network analysis. Compared with previous methods, the proposed method utilizes a spike transmission mechanism to simulate the propagation of action potentials in the brain, learning the coupling relationship between SC and FC, rather than solely focusing on the unified correlation between SC and FC topological patterns. Results on a real epilepsy dataset indicate that the proposed method not only outperforms existing methods across multiple evaluation metrics but also provides a new perspective for the study of SC-FC coupling.