Quantifying the Uncertainty of Activation Periods in fMRI Data via Changepoint Analysis

• Activity Number: 226
• Type: Topic Contributed
• Date/Time: Monday, July 30, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Statistics in Mental Health Research
• Author(s): Christopher Nam*+ and John Aston and Adam Johansen
• Companies: University of Warwick and University of Warwick and University of Warwick
• Keywords: changepoints ; fMRI ; segmentation ; hidden Markov models ; finite Markov chain imbedding ; sequential Monte Carlo

Abstract

Changepoint analysis has recently been used in fMRI data in determining activation periods of brain regions when the exact experimental design is unknown. This is particularly common in psychological experiments where different subjects react in different ways and different times to an equivalent stimulus. This talk will discuss a changepoint method which can consider multiple changepoints (activation periods), and quantify the uncertainty regarding these periods. The methodology combines recent work on evaluation of exact changepoint distributions conditional on model parameters via finite Markov chain imbedding in a hidden Markov model setting, and accounting for parameter uncertainty and estimation via Bayesian modelling and sequential Monte Carlo. This framework also allows for the inclusion of an autoregressive error process assumptions and detrending within a unified analysis. The methodology will be applied to fMRI data from a psychological experiment and demonstrate how detrending and error process assumptions can affect conclusions.