Abstract #0572

Resting-state fMRI Predicts Task Activation Patterns Using a Graph Convolutional Network

Zhangxuan Hu^1,2, Hua Guo², Lihong Wang³, Bing Wu¹, and Xue Zhang⁴

¹GE Healthcare, MR Research China, Beijing, China, ²Center for Biomedical Imaging Research, Department of Biomedical Engineering, School of Medicine, Tsinghua University, Beijing, China, ³Department of Psychiatry, University of Connecticut School of Medicine, Farmington, MI, United States, ⁴Department of Psychiatry and Behavioral Sciences, Stanford University, Stanford, CA, United States

Resting-state fMRI has the clinical potential as diagnostic and prognostic markers because of its easy implementation/standardization in data acquisitions, and its ability to parcellate functionally connected neural networks. It is of importance to examine whether the task-free spontaneous activity could be used to predict individuals’ task-induced activation. Here we proposed a graph convolutional network-based framework which utilized the information of the brain connections for the convolution step, and showed the ability of using resting-state fMRI to predict individual differences in activations of tasks from human connectome project. This framework could be extended to other resting-state fMRI researches.

This abstract and the presentation materials are available to members only; a login is required.

Join Here