Detecting changes in functional time series with applications to fMRI imaging

John Aston (University of Warwick)

Friday 16th March, 2012 15:00-16:00 Maths 203

Abstract

Change point detection in sequences of functional data is examined where the spatial functional observations are temporally dependent and where the distributions of change points from multiple subjects is required. Of particular interest is the case where the change point is an epidemic change (a change occurs and then the observations return to baseline at a later time). The special case where the covariance can be decomposed as a tensor product is considered with particular attention to the power analysis for detection. This is of interest in the application to functional magnetic resonance imaging (fMRI), where the estimation of a full covariance structure for the three-dimensional image is not computationally feasible. It is found that use of basis projections such as principal components for detection of the change points can be optimal in situations where PCA is traditionally thought to perform badly. This allows the investigation of "resting-state" fMRi, an experimental task traditionally assumed to be stationary and hence devoid of change points.[Joint work with Claudia Kirch, Karlsruhe Institute of Technology]     

Add to your calendar

Download event information as iCalendar file (only this event)