George L. Wilcox, Fellow

Axonal Propagation of Impulses in Peripheral Afferents

This research project investigates the mechanisms of pain signaling and brain processing. Computational work in the laboratory involves the simulation of electrical excitability using molecular biological data and the analysis of time series data emanating from the brain. An object-oriented approach to incorporating molecular information allows the researchers to simulate phenotypes or the effects of genetic variations and mutations at the tissue and organ level. The response of cells and networks to sensory input and pharmaceutical interventions is also being explored.

The researchers are using a sound numerical time-integration scheme comprising a backwarddifferential time-stepper, a modified Newton nonlinear solver, and a preconditioned Krylov subspace iterative linear solver. They are simulating A- δ and C-fibers from rat peripheral nerve because of molecular information available from collaborators in the Ion Channel Group of the Center for Mechanisms of Human Toxicity at the University of Leicester. The tools developed in this project have a wide range of applicability, serving as a prototype for exploring disturbance in electrical activity in any excitable cell in the body. The use of molecular information in simulations is a novel approach to understanding cell function. Mutation information will allow researchers to examine various diseased states and the use of pharmaceutical agents will allow them to explore options in patient therapy.

Further research by this group involves the development of parallelized neural network-based algorithms for spectral and statistical analysis of spatio-temporal data from the new magnetocephalography unit recently installed at the Brain Sciences Center (VA) in Minneapolis. The aims of this work are to develop new strategies of multidimensional time-series analysis methods using novel means of audio-visualization. Several approaches involving neural networks, information theory, cost function minimization, techniques based on spectral and components analyses, and new experimental concepts being developed in auditory data display are being applied towards the audio-visualization and analyses of these complex multidimensional data. The heavy volume and high dimensionality of the time-series data acquired simultaneously from multiple recording channels make the audio-visualization problem ideally suited to a parallel computing environment.