Decoding Functional Networks for Visual Categories via GNNs

2026-03-30

AI Summary

This study addresses the decoding of visual category-specific functional connectivity patterns from large-scale brain networks to elucidate the relationship between perception and cortical organization. Leveraging 7T fMRI data, the authors construct region-level functional connectivity maps and introduce, for the first time, a signed graph neural network (Signed GNN) to model both positive and negative connections. By integrating sparse edge masking with category-specific saliency analysis, the method effectively identifies functional network states associated with categories such as sports, food, and vehicles. Moving beyond traditional voxel-wise selectivity representations, this approach reveals reproducible, neuroscientifically interpretable subnetworks within the ventral and dorsal visual streams, offering a novel paradigm for understanding the network mechanisms underlying high-level visual processing.

Technology Category

Cognitive Modeling & Cognitive Systems: Neural Spike CodingComputer Vision: Visual Reasoning & Symbolic RepresentationsMachine Learning: Graph-based Machine Learning

Application Category

Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphsSocial Networks and Social Media: Social media analysis through the lenses of networksWeb Mining and Content Analysis: Web data visualization

Abstract

Understanding how large-scale brain networks represent visual categories is fundamental to linking perception and cortical organization. Using high-resolution 7T fMRI from the Natural Scenes Dataset, we construct parcel-level functional graphs and train a signed Graph Neural Network that models both positive and negative interactions, with a sparse edge mask and class-specific saliency. The model accurately decodes category-specific functional connectivity states (sports, food, vehicles) and reveals reproducible, biologically meaningful subnetworks along the ventral and dorsal visual pathways. This framework bridges machine learning and neuroscience by extending voxel-level category selectivity to a connectivity-based representation of visual processing.

Problem

• visual categories
• functional networks
• brain connectivity
• fMRI
• neural representation

Innovation

• signed Graph Neural Network
• functional connectivity
• visual category decoding
• 7T fMRI
• parcel-level brain graphs