Testing for Two Signals with Unknown Locations in a Gaussian Random Field: A Monte Carlo Simulation Study with an Application to fMRI

• Activity Number: 506
• Type: Contributed
• Date/Time: Wednesday, August 6, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Section on Medical Devices and Diagnostics
• Author(s): Pei-Chin Lu*+ and Khalil Shafie
• Companies: University of Northern Colorado and University of Northern Colorado
• Keywords: Gaussian random field ; fMRI ; Scale space

Abstract

Gaussian random field theory has been used extensively for correcting the multiple comparisons problem in neuroimaging over the past few decades. Siegmund and Worsley (1995) proposed the scale-space random fields for testing one signal with unknown location and scale. They showed that the global maximum of a Gaussian random field is the likelihood ratio test statistic in a N+1 dimensional "scale space", with N dimensions for location and 1 dimension for the width of a smoothing kernel. Nevertheless, it is not unusual to test for more than one signal in practice for an fMRI study. The current work is intended to extend Siegmund and Worsley's (1995) work under situations when the number of signals is two. Monte Carlo simulation will be used to examine the behaviors of the global maximum as well as to obtain the associated P-values.