Network Differential in Gaussian Graphical Models from Multimodal Neuroimaging Data.

Area of Science:

• Neuroscience
• Computational Psychiatry
• Network Science

Background:

• Multimodal brain network analysis offers potential for understanding brain disorders.
• Previous studies often focused on unimodal data or limited graph metrics, neglecting disrupted path details.
• Analyzing disrupted paths in multimodal brain graphs can reveal novel disease biomarkers.

Purpose of the Study:

• To develop a method for estimating multimodal brain graphs using static functional network connectivity (sFNC) and gray matter features.
• To identify path-based biomarkers in schizophrenia by analyzing disrupted network paths.
• To highlight the importance of multimodal analysis and path-based metrics for understanding brain disorders.

Main Methods:

• Estimated multimodal brain graphs using a Gaussian graphical model with sFNC and gray matter data from schizophrenia patients and controls.
• Applied graph theory to identify "disconnectors" or "connectors" in the patient graph, indicating altered paths compared to controls.
• Investigated disrupted paths within and between functional connectivity and gray matter networks.

Main Results:

• Identified specific edges in the schizophrenia graph associated with missing or additional paths compared to controls.
• Disrupted paths involved alterations both within and between functional connectivity and gray matter networks.
• Clinical relevance: A path-based biomarker was identified, with cross-modal edges linked to the middle temporal gyrus and cerebellum.

Conclusions:

• Multimodal brain network analysis combined with path-based disruption identification offers a more comprehensive understanding of schizophrenia.
• This approach can reveal path-based biomarkers that are missed by traditional pairwise edge analyses.
• The findings underscore the significance of integrating different data modalities and focusing on network path alterations for disease biomarker discovery.