Revealing Alzheimer's Disease Dementia Patterns in [18F]Florbetapir PET with Independent Component Analysis

This study investigates Alzheimer's Disease (AD) dementia through [18F]Florbetapir ([18F]FBP) Positron Emission Tomography (PET) imaging. We employ Independent Component Analysis (ICA) to identify shared latent patterns across controls and individuals with Dementia. The dataset comprises PET brain images from 440 participants from the Alzheimer's Disease Neuroimaging Initiative (ADNI). After visual inspection, nine independent components (IC) were selected, including visual, salience, default mode, cerebellum, left and right temporal, motor, frontal, and subcortical/brainstem. A Generalized Linear Model (GLM) analysis was performed on the IC weights to evaluate group differences. Salience, default mode, left and right temporal, and frontal components displayed a significant group effect with increased weights in the AD dementia group. Notably, the salience and frontal components demonstrated a significant interaction effect of diagnosis with age. This study emphasizes the potential of ICA in conjunction with [18F]FBP PET imaging to provide valuable insights into the neurobiology of AD dementia.