Effects of Brain Atlases and Machine Learning Methods on the Discrimination of Schizophrenia Patients: A Multimodal MRI Study

• DOI: 10.3389/fnins.2021.697168
• Publication Date: 2021-07-27T07:43:46Z
• Authors: Jinyu Zang; Yuanyuan Huang; Lingyin Kong; Bingye Lei; Pengfei Ke; Hehua Li; Jing Zhou; Dongsheng Xiong; Guixiang Li; Jun Chen; Xiaobo Li; Zhiming Xiang; Yuping Ning; Fengchun Wu; Kai Wu
• Keywords: Discriminative model; Hyperparameter

Recently, machine learning techniques have been widely applied in discriminative studies of schizophrenia (SZ) patients with multimodal magnetic resonance imaging (MRI); however, the effects brain atlases and methods remain largely unknown. In this study, we collected MRI data for 61 first-episode SZ (FESZ), 79 chronic (CSZ) 205 normal controls (NC) calculated 4 measurements, including regional gray matter volume (GMV), homogeneity (ReHo), amplitude low-frequency fluctuation degree centrality. We systematically analyzed performance two classifications (SZ vs NC; FESZ CSZ) based on combinations three atlases, five classifiers, cross validation 3 dimensionality reduction algorithms. Our results showed that groupwise whole-brain atlas 268 ROIs outperformed other atlases. addition, leave-one-out was best method to select hyperparameter set, but classification performances by different classifiers algorithms were quite similar. Importantly, contributions input features both higher GMV ReHo regions prefrontal temporal gyri. Furthermore, an ensemble performed establish integrated model, which improved. Taken together, these findings indicated factors constructing effective psychiatric diseases model has potential improve clinical diagnosis treatment evaluation SZ.