Exploring Brain Regions Related to Azheimer's Disease Using Functional Data Analysis Approach on Resting-State fMRI Data

Exploring Brain Regions Related to Azheimer's Disease Using Functional Data Analysis Approach on Resting-State fMRI Data

Alzheimer's Disease (AD) is a burdensome and incurable neurodegenerative disease, so brain study is needed to understand its pathogenesis. Since functional connectivity (FC) alteration has been shown in among AD patients groups, Resting-State fMRI (RS-fMRI) was used to explore brain regions. However, the previous study overlooked continuity on brain regions in FC analysis. From a study who introduced euclidean distance in FC analysis, we also employed a similar method and obtained FC curves. Since these curves can be considered to be functions, functional data analysis (FDA) was employed to treat this type of data. Especially, functional principal component analysis (FPCA) was used to extract function PC (fPC) scores from each brain region. These scores were used as covariates in logistic regression models with group penalties. Compared to the former approaches, our method showed higher classification rate in normal controls and AD patients (AUC = 0.9). Also, brain regions related to neurodegenerative studies were chosen in our best model. However, there were cases where only one side of the region was chosen. Hence further study is needed on these regions in our future work.