Automated, Robust Analysis of Quantitative Image Data Using Functional Mixed Models and Isomorphic Basis-Space Modeling

• Activity Number: 52
• Type: Invited
• Date/Time: Sunday, August 1, 2010 : 4:00 PM to 5:50 PM
• Sponsor: Section on Statistical Computing
• Author(s): Jeffrey S. Morris*+ and Veera Baladandayuthapani and Hongxiao Zhu
• Companies: MD Anderson Cancer Center and MD Anderson Cancer Center and MD Anderson Cancer Center
• Keywords: functional data analysis ; quantitative image analysis ; isomorphic transforms ; false discovery rate ; robust regression ; proteomics

Abstract

In this talk, I will describe flexible new automated methods to analyze quantitative image data. The methods are based on functional mixed models, a framework that can simultaneously model multiple factors and account for between-image correlation. We use an isomorphic basis-space approach to fitting the model, which leads to efficient calculations and adaptive smoothing yet flexibly accommodates the complex features characterizing these data. The method is automated and produces inferential plots indicating image regions associated with each factor, simultaneously considering practical and statistical significance, and controlling the false discovery rate. The method can be made robust to accommodate outlying images and handle heavy-tailed errors. It is applied to proteomic data, and is able to find results that would have been missed by conventional analysis approaches.