A potential diagnostic biomarker for schizophrenia based on local functional connectivity using dynamic regional
Lizhao Du1, Hongna Huang2, Zhengping Pu2
1Shanghai Med-X Engineering Research Center, School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai, China; Shanghai Mental Health Center, Shanghai Jiao Tong University School of Medicine, Shanghai, China; Shanghai Key Laboratory of Psychotic Disorders, Shanghai, China.
Objectives:
Though schizophrenia (SZ) has the well-established diagnostic criteria, the clinical conundrum of diagnostic inaccuracies still exists for its symptomatic overlap with other mental diseases like bipolar disorder (BD). Researchers have been looking for more specific and objective neuroimaging markers for SZ.
Methods:
Functional magnetic resonance imaging (fMRI) and T1 data from a total of 931 participants (SZ: 300; BD: 145; and healthy controls (HC): 486) were collected from two centers. Dynamic regional phase synchrony (DRePS) of BOLD signals were analyzed, as a potential discriminator from both HC and BD. Support vector machine (SVM) model, trained and tested for classifying SZ from HC in one center, was applied directly to external independent dataset. The same model was also trained and tested for classifying the SZ and BD subjects.
Results:
We found significant reduction of DRePS in SZ, compared with HC. There were also significant differences in DRePS between SZ and BD. Correlation analysis further showed prognostic value of DRePS for clinical behavior scoring (PANSS) (Spearman's ρ = 0.235, N = 166, p = .002, 95 % CI: [0.081, 0.378]). SVM model could obtain mean accuracies of 85 % and 72 % for classifying SZ from HC in the training center and the external center, respectively. When used for separating SZ and BD, SVM model could distinguish SZ from BD with mean accuracy ~72 %.
Conclusion:
DRePS of BOLD signals, which is correlated with the clinical symptoms, could be a potential neuroimaging biomarker separating SZ from both HC and BD.
