Whole brain atrophy based on iterative principal component analysis and MRI techniques in the study of Alzheimer's disease

Magnetic resonance imaging (MRI) based whole brain atrophy has been proposed as an imaging marker in clinical trials to evaluate the effects of potential treatments for Alzheimer's disease (AD) due to its objectiveness and sensitivity confirmed by a number of studies. Our study uses an iterative principal component analysis (IPCA) technique to measure whole brain atrophy from sequential MRI scans from baseline and 12month followup. IPCA-detected whole brain atrophy was significantly different among AD, MCI and normal controls (ANOVA p=2.3e-7, linear trend 5.9e-8 using disease severity of 0, 1, 2 for NC, MCI and AD). The IPCA-detected atrophy was highly correlated with the well-established brain boundary shift integration (BBSI) estimated whole brain atrophy (R=0.69, p=5.8e-6 for AD, R=0.53, p=1.2e-6 for MCI and R=0.3 and p=0.06 for NC). We conclude that IPCA based whole brain atrophy measure can be used together with MRI technique to distinguish patients with AD from normal controls and from MCI, and to potentially accelerate the treatment efficacy evaluation for AD.