Background: Respective changes in resting-state linear and nonlinear measures in major depressive disorder (MDD) have been reported. However, few studies have used integrated measures of linear and nonlinear brain dynamics to explore the pathological mechanisms underlying MDD. Method: Forty-two patients with MDD and 42 sex- and age-matched healthy controls (HC) underwent resting-state functional magnetic resonance imaging to calculate multiscale entropy (MSE) and regional homogeneity (ReHo). The MSE-ReHo coupling of the whole gray matter and the MSE/ReHo ratio (the complexity of intensity homogeneity per unit time series) of each voxel were compared between the two groups. To evaluate the discriminative capacity of ratio features between patients with MDD and HC, we employed the support vector machine (SVM) learning method. Results: We observed that patients with MDD displayed increased MSE/ReHo ratio mainly in the orbitofrontal cortex, sensorimotor areas, and visual cortex. Moreover, significant correlations were observed between MSE/ReHo ratio and clinical indicators, including depression severity and cognitive function tests. The SVM model demonstrated high accuracy in differentiating patients with MDD from HC, highlighting the potential of the MSE/ReHo ratio as a diagnostic and prognostic tool. Conclusions: The aberrant MSE/ReHo ratio implicated the underlying mechanisms of depressive symptoms and cognitive impairment in patients with MDD. It may represent a critical state of the brain region, reflecting the degree of chaos and order in the brain region. Integrating linear and nonlinear combinations of brain signals holds promise for diagnosing psychiatric disorders.
Read full abstract