Related Experiment Video
Updated: Dec 27, 2025
07:12
Individualized rTMS Treatment for Depression using an fMRI-Based Targeting Method
Published on: August 2, 2021
4.0K
Predicting Individual Remission After Electroconvulsive Therapy Based on Structural Magnetic Resonance Imaging: A
Akihiro Takamiya, Kuo-Ching Liang1, Shiro Nishikata1
1From the Department of Neuropsychiatry, Keio University School of Medicine.
The Journal of ECT
|March 3, 2020
Summary
Machine learning models using structural magnetic resonance imaging (MRI) data significantly improved prediction of electroconvulsive therapy (ECT) remission in depressed patients, achieving 93% accuracy. Key predictors included brain structure volumes and clinical features like psychotic symptoms.
Area of Science:
- Neuroimaging
- Machine Learning
- Psychiatry
Background:
- Electroconvulsive therapy (ECT) is an effective treatment for severe depression.
- Predicting individual response to ECT remains a clinical challenge.
- Machine learning offers potential for identifying predictive biomarkers.
Purpose of the Study:
- To identify clinical and neuroimaging features that predict electroconvulsive therapy (ECT) response.
- To develop machine learning models for predicting ECT remission and symptom severity.
Main Methods:
- Structural magnetic resonance imaging (MRI) and clinical data from 27 depressed patients undergoing ECT were analyzed.
- Support vector machine and regression models were built using clinical data, MRI data, or both.
- Leave-one-out cross-validation identified consistently selected predictive features.
Main Results:
- Models incorporating MRI data significantly improved ECT remission prediction accuracy from 70% to 93% compared to clinical data alone.
- Key predictive features included volumes of the gyrus rectus, right anterior lateral temporal lobe, cuneus, third ventricle, psychotic features, and family history of mood disorder.
Conclusions:
- Pretreatment structural MRI data enhance the individual prediction of ECT remission.
- A small subset of neuroimaging and clinical features are crucial for predicting ECT response.
