tsta

فارسی

Home / سی و دومین کنفرانس ملی و دهمین کنفرانس بین المللی مهندسی زیست پزشکی ایران

Graph Convolutional Network–Based Surrogate Modeling for MRI-EEG Connectivity Analysis

Authors :

Arshia Rezaei¹ Bahareh Abbaszadeh²

1- Center of Excellence in Design, Robotics, and Automation (CEDRA), School of Mechanical Engineering, Sharif University of Technology, Tehran, Iran 2- Department of Mechanical Engineering, Najafabad Branch, Islamic Azad University, Najafabad, Iran

Keywords :

Graph Convolutional Network (GCN)،Magnetic Resonance Imaging (MRI)،Electroencephalography (EEG)،Brain Connectivity

Abstract :

Multimodal neuroimaging data, comprising Magnetic Resonance Imaging (MRI) and Electroencephalography (EEG), provide complementary spatial and temporal perspectives on neural dynamics. Conventional multimodal connectivity estimation, however, is computationally demanding and often limited by incomplete or noisy inputs. This study presents a Graph Convolutional Network (GCN)-based framework for efficient approximation of connectivity patterns across MRI and EEG. The model exploits graph-structured representations from one modality to predict connectivity maps of the other, enabling bidirectional MRI→EEG and EEG→MRI inference. By embedding spatial dependencies among brain regions, the framework preserves both topological and spectral characteristics while substantially reducing computational cost. Experiments on multimodal datasets demonstrate that the GCN reliably reproduces full-model outputs, offering a scalable and robust solution for real-time, data-limited brain connectivity analysis.