Classification of schizophrenia by intersubject correlation in functional connectome

Area of Science:

• Neuroscience
• Brain Imaging
• Psychiatry

Background:

• Functional connectomes are proposed as unique identifiers.
• Similarities within phenotypic groups may aid in distinguishing psychiatric conditions.

Purpose of the Study:

• To investigate if functional connectome features can differentiate schizophrenia (SCH) patients from healthy controls (HCs) and depression patients.
• To explore the utility of group-specific functional connectomes (GFC) for individual classification.

Main Methods:

• Analysis of resting-state functional magnetic resonance imaging (fMRI) data from SCH, depression patients, and HCs across three centers.
• Calculation of connectome similarity within and between groups.
• Leave-one-out cross-validation using GFC for classification.
• Evaluation of classification performance and generalizability across centers.

Main Results:

• Higher functional connectome similarity was observed between subjects within the same group (e.g., SCH-SCH) compared to different groups (e.g., HC-SCH).
• GFC-based classification achieved significant accuracy (75-77%) and AUC (81-86%) in discriminating SCH from HC or depression.
• Cross-center classification demonstrated good generalizability of the GFC approach.

Conclusions:

• Average functional connectomes contain group-specific biological features.
• GFC analysis holds promise as a tool for clinical diagnosis in schizophrenia.
• Increasing sample size is more beneficial than increasing temporal resolution for improving classification performance.