Disentangling the association between functional connectivity and genetics in Mild Cognitive Impairment via multivariate regression models

Abstract

The aim of this study is to analyse links between phenotype and genotype for patients with Mild Cognitive Impariment (MCI) and Cognitively Normal (CN) using regression based model. Data was collected from 177 subjects downloaded from The Alzheimer’s Disease Neuroimaging Initiative (ADNI) database out of which 82 MCI and 95 CN were considered. In this research work multiple noise reduction pipelines were tested on resting state functional MRI data to analyse functional connectivity (FC) and extract imaging features. The pipelines performance was compared and nuisance regression based pipeline was selected. After preprocessing of the data FC matrices were generated for all subjects. Feature reduction technique was implemented to summarize the FC matrices into 28 imaging features. Finally, Partial Least Squares (PLS) model was applied to imaging and genetic features in which correlations are examined. A permutation test was performed to evaluate the model significance (p < 0:05). PLS model with LASSO results were investigated and associations between the imaging and genetic features were reported.