Multi-objective optimal experimental designs in event-related fMRI study
by
Abhyuday Mandal
University of Georgia
Coauthors: Jason MH Kao, John Stufken, Nicole Lazar

Well planned experimental designs are crucial to successfully achieving statistical goals under psychological restrictions in functional magnetic resonance imaging (fMRI) studies. With sophisticated allocations of stimuli, researchers can gather valuable fMRI time series and acquire precise information about human brain activities. However, due to the nature of fMRI experiments, the underlying design space is very large and irregular. This makes it difficult to find an optimal design that simultaneously accomplishes various goals of a study and fulfills the scientific restrictions. In this article, we propose an efficient, knowledge-based algorithm in constructing an optimal or near-optimal design that simultaneously achieves statistical goals, circumvents psychological confounds, and fulfills customized requirements for the event-related fMRI study when one or more stimulus types are involved. This proposed algorithm is demonstrated to outperform existing methodologies in convergence rate, achieved design efficiency and computation time. Another advantage of our approach is the incorporation of the HRF discretization interval. This allows the underlying model to precisely capture all the HRF lag effects at any sampling time point. Design measurements are defined and a global normalization is employed to ensure comparability of the measurements and to render a consistent design criterion. We also propose a novel use of traditional factorial designs in avoiding psychological effects.

PDF

Date received: July 10, 2007