Identifying current and remitted major depressive disorder with the Hurst exponent: a comparative study on two automated anatomical labeling atlases

Abstract

Major depressive disorder (MDD) is a leading world-wide psychiatric disorder with high recurrence rate, therefore, it is desirable to identify current MDD (cMDD) and remitted MDD (rMDD) for their appropriate therapeutic interventions. In the study, 19 cMDD, 19 rMDD and 19 well-matched healthy controls (HC) were enrolled and scanned with the resting-state functional magnetic resonance imaging (rs-fMRI). The Hurst exponent (HE) of rs-fMRI in AAL-90 and AAL-1024 atlases were calculated and compared between groups. Then, a radial basis function (RBF) based support vector machine was proposed to identify every pair of the cMDD, rMDD and HC groups using the abnormal HE features, and a leave-one-out cross-validation was used to evaluate the classification performance. Applying the proposed method with AAL-1024 and AAL-90 atlas respectively, 87% and 84% subjects were correctly identified between cMDD and HC, 84% and 71% between rMDD and HC, and 89% and 74% between cMDD and rMDD. Our results indicated that the HE was an effective feature to distinguish cMDD and rMDD from HC, and the recognition performances with AAL-1024 parcellation were better than that with the conventional AAL-90 parcellation.

Machine learning in major depression: From classification to treatment outcome prediction

“…Almost all of the selected studies used SVM or its variant method as the primary classification method 22,[40][41][42][44][45][46][47][48][49][50]52,53,82,84,87,[89][90][91]98 and use LOOCV for cross validation. The reason why SVM is the most popular choice among depression classification is because of its useful strengths on including a reliable theoretical foundation and its flexible response to high-dimensional data.…”

A Brainnetome Atlas Based Mild Cognitive Impairment Identification Using Hurst Exponent

“…Different parcellation schemes may lead to different classification results. Compared to the widely used automated anatomical labeling atlas, the brainnetome atlas that simultaneously combines information from structural and functional connections obtained better classification performance in differentiating major depressive disorder from HC in our previous study (Jing et al, 2017 ). Thus the brainnetome atlas was adopted to discriminate MCI from HC subjects in this work.…”

“…The range scaled analysis, which is an effective method to detect the temporal complexity of a time series, was utilized to calculate the HE index of fMRI signals at a voxel level, and the detailed principle of HE calculation was reported in our previous study (Jing et al, 2017 ). In addition, the brainnetome atlas (Figure 1 ), which partitions the cerebral cortex into 246 ROIs including 210 cortical sub-regions and 36 subcortical sub-regions (Fan et al, 2016 ), was used to extract the HE index feature for the SVM-based classification algorithm.…”

“…In this case, the dynamics of a time series would keep a reversing pattern in time, and a decrease in the time series generally would be followed an increase and vice versa (Gentili et al, 2015 ). Recently, HE index has been utilized to measure the changes of BOLD signals related to major depressive disorder, normal and pathological aging, cholinergic modulation, AD, autism disorder and different personality traits (Maxim et al, 2005 ; Wink et al, 2006 ; Lai et al, 2010 ; Lei et al, 2013 ; Gentili et al, 2015 ; Jing et al, 2017 ). However, little information was known about the HE changes in MCI patients, and it still remains unknown whether the HE index could serve as an effective parameter for MCI classification.…”

Identifying resting‐state effective connectivity abnormalities in drug‐naïve major depressive disorder diagnosis via graph convolutional networks

“…A typical ML pipeline for the diagnosis of MDD can be summarized as follows: feature extraction, feature selection, model training, classification, and performance evaluation. In studies that differentiate MDD patients from healthy controls (HC), the following have been used as features extracted from rs‐fMRI: spatial independent components (Ramasubbu et al, 2016; Wei et al, 2013), the Hurst exponent (Jing et al, 2017), degree centrality (Li et al, 2017), and regional homogeneity (Ma, Li, Yu, He, & Li, 2013). In addition, many previous studies also applied graph theory approaches (Bhaumik et al, 2017; Cao et al, 2014; Drysdale et al, 2017; Guo et al, 2014; Lord, Horn, Breakspear, & Walter, 2012; Sundermann et al, 2017; Wang, Ren, & Zhang, 2017; Yoshida et al, 2017; Zeng, Shen, Liu, & Hu, 2014; Zhong et al, 2017) to the preestimated FC for investigating the disrupted functional brain networks in MDD patients.…”