Structural connectivity alterations in mild cognitive impairment patients with depression symptoms: a DTI-based connectome analysis

Abstract

Abstract Studies have shown that depressive symptoms cause changes in brain structural network, but the characteristics of brain structural network in mild cognitive impairment with depression symptoms (D-MCI) are not well understood. In this study, we used diffusion tensor imaging and graph theory analysis to investigate abnormalities in brain structural networks in mild cognitive impairment with depression symptoms. We acquired magnetic resonance imaging data from 50 subjects on a 3T MRI. Subjects collected included 14 patients with D-MCI, 18 patients with MCI with no depression (nD-MCI), and 18 healthy controls. We utilized the network-based statistics method to explore the changes in the structural networks between the three groups and the classification capabilities combined with machine learning methods. In contrast to healthy controls, the anomalous subnetworks of MCI revealed by network-based statistics are mainly located in the default mode network, basal ganglia and sensorimotor regions. The classification accuracy of machine learning models is D-MCI vs nD-MCI: 77.5%; D-MCI vs healthy controls: 90.0%; nD-MCI vs healthy controls: 86.7%. Our results suggest that depressive symptoms cause changes in structural network in patients with MCI, and that these changes can be used to distinguish between D-MCI, nD-MCI, and healthy controls.