Use of Parallel Simulated Annealing for Computational Modeling of Human Head Conductivity

Abstract

Abstract. We present a parallel computational environment used to determine conductivity properties of human head tissues when the effects of skull inhomogeneities are modeled. The environment employs a parallel simulated annealing algorithm to overcome poor convergence rates of the simplex method for larger numbers of head tissues required for accurate modeling of electromagnetic dynamics of brain function. To properly account for skull inhomogeneities, parcellation of skull parts is necessary. The multi-level parallel simulated annealing algorithm is described and performance results presented. Significant improvements in both convergence rate and speedup are achieved. The simulated annealing algorithm was successful in extracting conductivity values for up to thirteen head tissues without showing computational deficiency.

Skull Modeling Effects in Conductivity Estimates Using Parametric Electrical Impedance Tomography

“…The process of fitting the bEIT data and the simulations is a nonlinear optimization problem, known as the bEIT inverse problem (IP). The methods for solving the IP include Newton's method [35], simplex search [16], [45], least squares through a linearization [46] and simulated annealing [47], [48], among others.…”

Next-generation human brain neuroimaging and the role of high-performance computing

“…Then the response is measured on the other electrodes. Once an appropriate objective function describing the difference between the measured scalp potentials, V , and the predicted potentials φ p , is defined (e.g., least square norm), a search for the global minimum is undertaken using nonlinear optimization algorithms (e.g., simulated annealing [42], [43] or simplex search). Using either optimization method, the search for the optimal conductivities requires a large number of forward calculations, in the order of 3K for a single current injection pair.…”

“…Depending on the number of conductivity unknowns, each conductivity search for a single pair will require many thousands of forward solutions to be generated. Simulated annealing is currently used as the optimization strategy [43] and our parallel implementation will support up to twelve simultaneous forward solves. Clearly, conductivity results for each pair can also be done in parallel.…”

An Open Domain-Extensible Environment for Simulation-Based Scientific Investigation (ODESSI)

“…ODESSI was inspired by our research in human neuroscience where we are developing computational models of human head electromagnetics for use in dynamic brain analysis [18,20,21]. The main goal of our research is to estimate the locations of the active brain regions given measured electroencephalogram (EEG) recordings.…”

“…The development of the ODESSI prototype is discussed in §4. ODESSI was inspired by our prior ICCS work [18,20] on computational modeling of human head conductivity. Section §5 outlines the domain problem in human brain science we are investigating and shows how ODESSI is applied to improve scientific productivity in this domain.…”