Multivariate Classification of Major Depressive Disorder Using the Effective Connectivity and Functional Connectivity

Area of Science:

• Neuroscience
• Psychiatry
• Medical Imaging

Background:

• Major depressive disorder (MDD) diagnosis is challenging using rest-state fMRI due to data complexity.
• Previous studies have not focused on classifying MDD using both effective and functional connectivity.

Purpose of the Study:

• To classify major depressive disorder (MDD) patients from healthy controls using whole-brain connectivity measures.
• To compare the diagnostic potential of effective connectivity versus functional connectivity in MDD.

Main Methods:

• Extracted effective connectivity using spectral Dynamic Causal Modeling (spDCM) and functional connectivity from resting-state fMRI.
• Utilized classifiers including linear SVM, non-linear SVM, KNN, and LR for data-driven classification.
• Analyzed connectivity within and across default mode network (DMN), dorsal attention network (DAN), frontal-parietal network (FPN), and silence network (SN).

Main Results:

• Effective connectivity achieved the highest accuracy of 91.67% using 19 connections.
• Functional connectivity achieved 89.36% accuracy using 6,650 connections.
• Discriminative effective connections were identified in regions like PCC, vmPFC, dACC, and IPL.

Conclusions:

• Effective connectivity measures demonstrate superior diagnostic potential for MDD compared to functional connectivity.
• These findings suggest effective connectivity offers better mechanistic interpretability for MDD.
• High accuracies indicate a diagnostic potential for earlier MDD prevention or intervention.