Schizophrenia Detection Based on Morphometry of Hippocampus and Amygdala

Area of Science:

• Neuroimaging
• Psychiatry
• Machine Learning

Background:

• Schizophrenia (SZ) is a severe mental disorder linked to brain abnormalities, particularly in the hippocampus and amygdala.
• Previous research focused on volume, but surface-based morphometry offers finer detail on deformation, yet its relation to SZ pathology and biomarker potential remains unclear.

Purpose of the Study:

• To investigate morphometric differences in the hippocampus and amygdala between individuals with schizophrenia and healthy controls using multivariate morphometry statistics (MMS).
• To evaluate the potential of MMS-derived features as biomarkers for schizophrenia diagnosis.

Main Methods:

• Extracted individual MMS of the hippocampus and amygdala from MRI images.
• Applied dictionary learning and max pooling for feature reduction, followed by machine learning for classification.
• Utilized a random forest classifier to assess diagnostic accuracy.

Main Results:

• Significant bilateral hippocampal atrophy, symmetrical in individuals with schizophrenia.
• Both atrophied and expanded subregions within the amygdala, with greater deformation in the right amygdala.
• High classification accuracies: 94.52% (hippocampus), 94.57% (amygdala), and 96.57% (combined features).

Conclusions:

• MMS effectively identifies morphometric differences in the hippocampus and amygdala between schizophrenia patients and controls.
• These findings highlight the potential of MMS as a reliable biomarker for schizophrenia diagnosis.