Functional Data Techniques for Mapping of Neurodevelopmental Trajectories

• Activity Number: 108
• Type: Invited
• Date/Time: Monday, August 5, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Philip Reiss*+ and Lei Huang and Huaihou Chen and Thaddeus Tarpey
• Companies: New York University and Johns Hopkins University and New York University and Wright State University
• Keywords: diffusion tensor imaging ; tensor product splines ; functional principal components ; brain development ; functional data clustering

Abstract

A major goal of neuroimaging research is to map trajectories of normal and abnormal development. In many applications we are given a set of curves or images derived from individuals of different ages, where each person's curve represents a quantity of neuroscientific interest measured along a set of brain locations. The objective is to fit the quantity of interest as a smooth function of age at each location, while appropriately borrowing strength across locations. The functional data paradigm can be useful here in at least two ways. (1) Formulating the problem as one of relating functional responses (the image-derived data) to a scalar predictor (age) motivates a notion of pointwise degrees of freedom that provides a unifying framework for some existing approaches, and also motivates several new ones. (2) Functional data clustering offers an improved way of mapping the results. The methods will be illustrated by application to a study of white matter microstructure in the corpus callosum.