Classifying Schizophrenia Patients and Healthy Individuals: Whole Brain SPECT Functional Connectivity using Support Vector Machine Classification

Background:Functional magnetic resonance imaging (fMRI) has been used to characterize functional brain networks in disorders ranging from depression to schizophrenia (Chatterjee & Mittal, 2019; Pilmeyer et al., 2022). Like fMRI, single photon emission computed tomography (SPECT) is a technique which captures information about neurally linked blood flow activity through radioactive tracers (Davis et al., 2020). While a few SPECT studies in schizophrenia populations have been conducted (Malaspina et al., 1999) along with fMRI based studies (Steardo et al., 2020), research on SPECT data for individual subject classification is limited. We first used an independent component analysis (ICA) approach to estimate covarying SPECT networks (Harikumar et al., 2025). Results were then fed as input to a classifier model to evaluate accuracy of individual diagnostic prediction.Methods:213 subjects (137 schizophrenia patients and 76 healthy controls) were used for the analysis. Classification input was based on loading parameters generated from spatially constrained ICA using a set of network priors derived from fMRI. Fifty-three SPECT components were estimated guided by the NeuroMark fMRI 1.0 template (Du et al., 2020). We initially focused on a support vector machine (SVM) classifier given previous favorable fMRI-SVM results. We also evaluated performance of multiple classifiers post hoc.Results and Conclusion:Linear SVM classification results showed a cross-validated classifier score (area under the curve (AUC)) of 83% (SD = 0.089), significantly above chance. Auditory, subcortical and sensorimotor networks were the highest ranked features. Results provide the first cross-validated classification of individual subject diagnoses using SPECT brain networks.