A graph neural network for brain functional connectivity analysis
Elena N. Pitsik, 1, Semen A. Kurkin, 1, Alexander E. Hramov, 2, Rositsa Paunova, 3, Denitsa Simeonova, 3, Sevdalina Kandilarova 3, Drozdstoy Stoyanov, 3
1 Samara State Medical University, Samara, Russia
2 Institute of Cardiological Research, Saratov State Medical University, Saratov, Russia
3 Plovdiv Medical University, Plovdiv, Bulgaria
Abstract
In the present paper, we propose a model based on graph neural network for analysis of brain functional connectivity of the patients with mood disorders. Graph neural networks are ideal for working with graph-like data due to their inherent property to consider the connectivity structure of the data. We demonstrate that the model achieves good generalisation ability on major depression disorder rs-fMRI despite the moderate amount of training data, and is able to select the subset of regions consistent with the existing MDD connectivity literature.
Speaker
Elena Pitsik
Innopolis University
Russia
Report
File with report
Discussion
Ask question