Fronto-Temporal Connectivity Predicts ECT Outcome in Major Depression.

Area of Science:

• Neuroscience
• Psychiatry
• Medical Imaging

Background:

• Electroconvulsive therapy (ECT) is a highly effective treatment for severe depression.
• Predicting ECT response using MRI data is an active area of research.
• Selecting optimal MRI metrics for prediction remains a challenge.

Purpose of the Study:

• To assess the predictive capability of aggregated functional MRI metrics for ECT antidepressant response.
• To utilize machine learning to analyze basal brain activity and connectivity patterns.
• To identify key neuroimaging features associated with ECT treatment outcomes.

Main Methods:

• Trained a radial support vector machine using arterial spin labeling (ASL) and BOLD fMRI metrics from 46 depressed patients.
• Included metrics such as cerebral blood flow, regional homogeneity, and graph theory measures.
• Employed a 5-repeated 5-fold cross-validation with nested feature selection for model validation.

Main Results:

• Models achieved balanced accuracy ranging from 58-68%, outperforming chance.
• Prediction accuracy was higher for non-responders (up to 74%) than responders (up to 64%).
• Consistently selected features included connectivity within fronto-parietal, motor, temporal networks, and the subgenual anterior cingulate cortex (sgACC).

Conclusions:

• Multimodal fMRI metrics can successfully predict ECT outcomes, especially for non-responders.
• Brain connectivity patterns involving the left dorsolateral prefrontal cortex (DLPFC) and sgACC are crucial predictive features.
• Findings suggest potential for refining neurostimulation targets and improving personalized depression treatment.