Application of a Hybrid Genetic/Powell Algorithm and a Boundary Element Method to Electrical Impedance Tomography

An optimization method based on a genetic algorithm (GA) and a boundary element method is applied to solve an electrical impedance tomography problem. The scheme is applied to reconstruct highly irregular shapes and to image and count objects inside a host medium of different impedance. A Pareto multiobjective optimization method is applied to improve the performance of the GA. Comparisons between the GA and a calculus-based method for selected test problems show that the calculus-based method outperforms the GA in simple cases but that for more complex cases the GA reaches the correct solution whereas the calculus-based method does not. A hybrid scheme that we developed combining a calculus-based method and the GA is shown to be the most efficient and robust even when applied to the complex cases we tested. The sensitivity of the current scheme is evaluated in the presence of noise.