Hybrid ICA-Bayesian network approach reveals distinct effective connectivity differences in schizophrenia

Area of Science:

• Neuroscience
• Computational Psychiatry
• Medical Imaging

Background:

• Schizophrenia is associated with altered brain connectivity.
• Effective connectivity analysis is crucial for understanding brain function in psychiatric disorders.
• fMRI data often contains noise that can impede connectivity analysis.

Purpose of the Study:

• To investigate differences in effective brain connectivity between schizophrenia patients and healthy controls.
• To evaluate the impact of independent component analysis (ICA) filtering on discrete dynamic Bayesian network (dDBN) analysis of fMRI data.
• To identify specific brain regions with altered connectivity in schizophrenia.

Main Methods:

• Utilized a discrete dynamic Bayesian network (dDBN) approach to analyze effective connectivity.
• Employed an independent component analysis (ICA) filtering method to reduce noise in fMRI data.
• Compared dDBN analysis on both filtered and unfiltered fMRI data from 35 schizophrenia patients and 35 healthy controls.

Main Results:

• ICA filtering significantly improved the magnitude of the approximate conditional likelihood (ACL) score, a measure of effective connectivity.
• Schizophrenia patients demonstrated significantly lower ACL scores compared to healthy controls.
• The most pronounced differences in connectivity were observed in the cerebellar vermis and hemispheres, as well as bilateral temporal and frontal regions.

Conclusions:

• Schizophrenia is characterized by weaker effective connectivity in key brain regions, including cerebellar, temporal, and frontal areas.
• ICA filtering enhances the reliability and sensitivity of dDBN analysis for fMRI data in psychiatric research.
• These findings contribute to a better understanding of the neural underpinnings of schizophrenia.