High-Dimensional Longitudinal Functional Principal Component Analysis for RAVENS Images

• Activity Number: 335
• Type: Contributed
• Date/Time: Tuesday, July 31, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Seonjoo Lee*+ and Brian Scott Caffo and Ciprian Crainiceanu and Vadim Zipunnikov and Dzung L Pham
• Companies: Henry Jackson Foundation and The Johns Hopkins University and The Johns Hopkins University and Johns Hopkins Bloomberg School of Public Health and The Henry M. Jackson Foundation
• Keywords: longitudinal functional principal component analysis ; structural brain images ; high-dimensional statistics ; voxel-based morphometry

Abstract

We describe and analyze longitudinal magnetic resonance imaging (MRI) data to reveal brain atrophy and ventricular enlargement using regional analysis of volumes examined in normalized space (RAVENS) and voxel-based morphometry (VBM) statistical analysis. Current longitudinal VBM analysis are limited in separating registration errors from longitudinal changes and revealing relationships between baseline patterns and longitudinal trends. To analyze the data, we propose a longitudinal functional principal component analysis (LFPCA) for high-dimensional images: we decompose RAVENS images measured at multiple visits into three additive components: a subject-specific functional random intercept that quantifies the cross-sectional variability, a subject-specific functional slope that quantifies the irreversible changes over multiple visits, and a subject-visit specific functional deviation. We describe strategies to identify baseline/longitudinal variation and registration errors combined with covariates of interest.