Disrupted putamen-centered circuitry in major depressive disorder: Evidence from multimodal neuroimaging and machine learning.
Authors
Affiliations (6)
Affiliations (6)
- Department of Radiology, Suzhou Hospital, Xiyuan Hospital of China Academy of Chinese Medical Sciences (Suzhou TCM Hospital), Suzhou, China.
- Beijing Anzhen Hospital Affiliated to Capital Medical University, Beijing, China.
- Department of Radiology, Affiliated Guangji Hospital of Soochow University, Suzhou, China.
- Department of Radiology, Yixing Hospital Affiliated to Jiangsu University, Wuxi, China.
- Yancheng Second People's Hospital, Yancheng, China.
- Department of Radiology, Suzhou TCM Hospital Affiliated to Nanjing University of Chinese Medicine, Suzhou, China. Electronic address: [email protected].
Abstract
The pathophysiology of Major Depressive Disorder (MDD) extends beyond focal limbic regions to involve distributed neural circuits, yet multimodal alterations in subcortical networks remain underexplored. This study aims to characterize this dysfunction and evaluate the diagnostic potential of neuroimaging features. We analyzed multimodal MRI data from MDD patients and matched healthy controls. Subcortical volumetric and functional connectivity (FC) features trained an ExtraTrees classifier with nested cross-validation for patient/control discrimination. MDD patients exhibited co-occurring volume reductions in the right thalamus and left putamen, along with reduced FC in two putamen-centered pathways (right putamen-left putamen and right putamen-left amygdala). A positive structure-function coupling was observed between left putamen volume and its interhemispheric FC. The multimodal model achieved the highest classification performance (AUC = 0.685, balanced accuracy = 65.3%), outperforming both FC-only (AUC = 0.655) and volume-only models (AUC = 0.649). Feature importance analysis identified the two putamen-centered FC connections as the top discriminative predictors. These findings indicate that putamen-centered functional dysconnectivity represents a stable trait-like neural signature of MDD, contributing to classification despite lacking correlation with symptom severity, supporting MDD as a disorder of large-scale brain network integration.