Related Experiment Video
Updated: Jul 28, 2026
12:09
Network Analysis of the Default Mode Network Using Functional Connectivity MRI in Temporal Lobe Epilepsy
Published on: August 5, 2014
18.2K
Multi-atlas ensemble graph neural network model for major depressive disorder detection using functional MRI data
Nojod M Alotaibi1, Areej M Alhothali1, Manar S Ali1
1Department of Computer Science, Faculty of Computing and Information Technology, King Abdulaziz University, Jeddah, Saudi Arabia.
Frontiers in Computational Neuroscience
|June 24, 2025
Summary
This study introduces a novel ensemble graph neural network (GNN) model using functional MRI data to improve major depressive disorder (MDD) detection. The model shows promise for accurate and generalizable diagnosis of MDD.
Area of Science:
- Neuroscience
- Artificial Intelligence
- Medical Imaging
Background:
- Major Depressive Disorder (MDD) is a leading cause of disability, often diagnosed through subjective symptoms, overlooking complex pathophysiology.
- Current diagnostic methods for MDD lack objective measures, highlighting the need for advanced techniques to understand its brain network disorder nature.
- Neuroimaging, particularly rest-state functional MRI (rs-fMRI), is crucial for investigating brain alterations in MDD.
Purpose of the Study:
- To develop an ensemble-based Graph Neural Network (GNN) model for enhanced detection of Major Depressive Disorder (MDD) using rs-fMRI data.
- To leverage multi-atlas functional connectivity features to capture brain complexity and improve diagnostic accuracy for MDD.
- To validate the model's generalizability and reliability on a large, multi-site MDD dataset.
Main Methods:
- Utilized rest-state functional Magnetic Resonance Imaging (rs-fMRI) data from a large, multi-site cohort diagnosed with MDD.
- Developed an ensemble Graph Neural Network (GNN) model integrating functional connectivity derived from multiple brain region segmentation atlases.
- Applied Synthetic Minority Over-sampling Technique (SMOTE) for class imbalance and employed stratified 10-fold cross-validation for performance assessment.
Main Results:
- The proposed multi-atlas ensemble GNN model achieved a high accuracy of 75.80% in detecting MDD.
- Key performance metrics included sensitivity of 88.89%, specificity of 61.84%, precision of 71.29%, and an F1-score of 79.12%.
- The ensemble approach demonstrated superior performance compared to single-atlas models in capturing discriminative features for MDD diagnosis.
Conclusions:
- The developed ensemble GNN model offers a reliable and generalizable approach for the accurate detection of MDD from rs-fMRI data.
- This study underscores the potential of deep learning techniques, specifically GNNs, in advancing the objective diagnosis of mental health disorders.
- Integrating multi-atlas functional connectivity enhances the model's ability to identify complex brain network alterations characteristic of MDD.
