A three-way parallel ICA approach to analyze links among genetics, brain structure and brain function

Area of Science:

• Neuroscience
• Medical Imaging
• Genetics
• Biomedical Data Fusion

Background:

• Multi-modal data analysis is crucial in neuroscience, medical imaging, and genetics.
• Parallel Independent Component Analysis (pICA) enables simultaneous decomposition of two data modalities, revealing inter-modal links.
• Increasing data acquisition yields more than two modalities per subject, necessitating advanced analysis techniques.

Purpose of the Study:

• To extend the pICA approach to analyze three data modalities simultaneously.
• To evaluate the performance of three-way pICA in identifying pairwise links and estimating sources.
• To apply the three-way pICA algorithm to real-world biomedical data for investigating genetic effects on alcohol dependence.

Main Methods:

• Development and simulation of a three-way Parallel Independent Component Analysis (pICA) algorithm.
• Application of the three-way pICA to functional MRI, structural MRI (gray matter concentration), and genetic (SNP) data.
• Comparison of three-way pICA performance against standard pICA and separate ICA analyses.

Main Results:

• Simulations demonstrated that three-way pICA accurately identifies pairwise links and estimates sources better than existing methods.
• The algorithm successfully linked a SNP component (associated with BDNF, GRIN2B, NRG1) to functional and structural brain components in alcohol dependence data.
• Identified links included precuneus activation (functional) and default mode network/caudate regions (gray matter).

Conclusions:

• The three-way pICA algorithm is a validated tool for comprehensive multi-modal biomedical data fusion.
• The approach offers improved performance over existing methods for analyzing three or more data modalities.
• Findings suggest potential genetic influences on brain structure and function related to alcohol dependence, warranting further investigation.