Identification of Quantitative Susceptibility Biomarkers for First Episode Psychosis using XGBoost

Several studies have demonstrated altered neurochemicals in deep-brain nuclei of psychosis patients. These alterations suggest a dopamine dysfunction in subcortical areas. QSM images quantify magnetic susceptibility changes in the brain. These changes are usually associated with iron concentrations and co-factor in dopamine pathways of the neurons. We propose a method based on machine learning to discriminate between psychosis patients and healthy controls by looking at the magnetic susceptibility of 9 deep gray matter nuclei, obtaining a precision of 91.6%.