Structural brain networks in schizophrenia based on nonnegative matrix factorization.

Area of Science:

• Neuroscience
• Psychiatry
• Brain Imaging

Background:

• Schizophrenia is a severe mental disorder characterized by significant gray matter volume reductions and cortical thickness alterations in various brain regions.
• Previous research primarily focused on voxel-level changes, often overlooking the spatial relationships and network-level alterations within the brain.

Purpose of the Study:

• To investigate structural covariance at the network level in individuals with schizophrenia using a novel data-driven technique.
• To identify specific brain networks affected by gray matter volume reductions in schizophrenia.

Main Methods:

• Employed nonnegative matrix factorization (NMF), a data-driven method, to group voxels with similar information into structural covariance networks.
• Analyzed brain morphometry data from 36 patients with schizophrenia and 21 healthy controls.

Main Results:

• Patients with schizophrenia demonstrated significant gray matter volume reductions in six identified structural covariance networks.
• Affected networks included the dorsal striatum, thalamus, hippocampus-parahippocampus, supplementary motor area-fusiform, middle/inferior temporal network, and frontal-parietal-occipital network.
• Findings support the hypothesis of disrupted cortical-subcortical circuits in schizophrenia.

Conclusions:

• Nonnegative matrix factorization (NMF) is an effective multivariate method for identifying brain networks and studying structural covariance.
• This network-level approach offers a new perspective for investigating the neural mechanisms underlying schizophrenia.