Discovering robust biomarkers of psychiatric disorders from resting-state functional MRI via graph neural networks: A systematic review

Published: 01 Jan 2025, Last Modified: 04 Oct 2025NeuroImage 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•A wide range of predictors and attributors have been used in existing fMRI studies.•Robustness of salient features highlighted by attributors remains underexplored.•We propose a taxonomy to classify state-of-the-art GNNs customised to fMRI datasets.•We propose a framework to evaluate potential biomarkers derived via machine learning.•Some key features at the level of regions of interest were consistent across studies.
Loading