Parameter Sweeps for Functional MRI Research in the "Virtual Laboratory for e-Science" Project

Abstract

Image analysis is an important component of neuroscience research. The ICT infrastructure and technical knowledge needed to perform (large scale) neuroimaging studies, however, is often not available to the neuroscientists. The "virtual laboratory for e-sciences" project provides an advanced (grid) infrastructure offering data and computing services to researchers from several application domains. In this paper we describe how this infrastructure is used in the context of functional magnetic resonance imaging (MRI) research, which is devoted to the study of brain activity due to stimulation. Our experience in using a generic application (Nimrod, Monash University, Australia) to manage large parameter sweep experiments is presented. These experiments were performed to investigate the effect of one parameter (delay in the hemodynamic response function) in the analysis result, but also to evaluate the available infrastructure (grid resources and experiment management services). Initial results indicate that it is feasible and simple to perform large image analysis experiments using Nimrod once the environment has been properly prepared.