Identify connectome between genotypes and brain network phenotypes via deep self-reconstruction sparse canonical...

Area of Science:

• Neuroscience
• Genetics
• Medical Imaging

Background:

• Brain imaging genetics is an emerging field investigating the genetic basis of brain structure and function.
• Identifying which brain imaging phenotypes are most influenced by genetic effects remains a challenge.

Purpose of the Study:

• To develop a method for correlating genetic variants (single nucleotide polymorphisms, SNPs) with brain network quantitative traits (QTs).
• To identify the connectome, including brain regions and connectivity features, from functional magnetic resonance imaging (fMRI) data.
• To discover genetic associations with functional connectivity and brain region phenotypic biomarkers.

Main Methods:

• Constructed a connection matrix from fMRI data, selecting upper triangle elements as connectivity features.
• Utilized the PageRank algorithm to determine brain region importance as features.
• Developed a deep self-reconstruction sparse canonical correlation analysis (DS-SCCA) method for multi-SNP-multi-QT association analysis.
• Optimized the DS-SCCA method using parametric approaches, augmented Lagrange, and stochastic gradient descent.

Main Results:

• The DS-SCCA approach successfully identified strong genetic associations with functional connectivity and brain region phenotypes.
• Demonstrated the method's scalability and suitability for large datasets like the Alzheimer's Disease Neuroimaging Initiative (ADNI).
• Discovered phenotypic biomarkers that can guide disease interpretation.

Conclusions:

• The developed DS-SCCA method provides a powerful tool for imaging genetic association studies.
• This approach advances the understanding of the genetic architecture of brain networks.
• The identified biomarkers hold potential for future research in neurodegenerative diseases.