fMRI Design under Autoregressive Model with One Type of Stimulus

abstract: Functional magnetic resonance imaging (fMRI) is used to study brain activity due to stimuli presented to subjects in a scanner. It is important to conduct statistical inference on such time series fMRI data obtained. It is also important to select optimal designs for practical experiments. Design selection under autoregressive models have not been thoroughly discussed before. This paper derives general information matrices for orthogonal designs under autoregressive model with an arbitrary number of correlation coefficients. We further provide the minimum trace of orthogonal circulant designs under AR(1) model, which is used as a criterion to compare practical designs such as M-sequence designs and circulant (almost) orthogonal array designs. We also explore optimal designs under AR(2) model. In practice, types of stimuli can be more than one, but in this paper we only consider the simplest situation with only one type of stimuli.Dissertation/ThesisMasters Thesis Statistics 201