Discriminating subclinical depression from major depression using multi-scale brain functional features: A radiomics...

Area of Science:

• Neuroimaging
• Radiomics
• Machine Learning in Medicine

Background:

• Current diagnosis of subclinical depression (SD) and major depression (MD) relies on subjective methods, increasing misdiagnosis risk.
• Objective diagnostic markers are needed to differentiate SD from MD accurately.

Purpose of the Study:

• To develop and validate a radiomics-based classification method for distinguishing SD from MD using resting-state functional magnetic resonance imaging (rs-fMRI) data.
• To identify key functional brain features that differentiate SD and MD.

Main Methods:

• rs-fMRI data acquired from 26 SD, 36 MD subjects, and 33 healthy controls (HC).
• Novel radiomics analysis involving multi-scale functional feature extraction.
• Two-level feature selection and Support Vector Machine (SVM) for classification.

Main Results:

• The classification model achieved an overall accuracy of 84.21% for SD, MD, and HC groups.
• Excellent discrimination between SD and MD with 96.77% accuracy, 100% sensitivity, and 92.31% specificity.
• Key differentiating features were associated with default mode, frontoparietal, affective, and visual networks.

Conclusions:

• A radiomics approach using functional brain measures can effectively and objectively differentiate SD from MD.
• This method holds promise for improving individual clinical diagnosis and understanding the neurobiological underpinnings of SD and MD.