Calculation of reliability maps with respect to inaccuracy in fMRI data

Abstract

In functional MRI, accurate knowledge of experimental stimulus time-course, which depends on accuracy of subject's behavioral response, is necessary to obtain valid and meaningful results. For one type of inaccuracy, missing response to stimulus, we present the methods to calculate reliability maps. Two criterions to calculate maximal tolerable inaccuracy are proposed. Their combination can guarantee both detection of activation for inaccurate model and insignificant difference between accurate and inaccurate results. We can use the maps to find weak and/or robust activations in accurate realization of experiment, and subsequently use it to prevent us from wrong inferences in inaccurate realization of experiment.