Numerical optimization of 3-D SAR distributions in cylindrical models for electromagnetic hyperthermia.

ABSTRACT

Numerically optimized SAR (specific absorption rate) distributions in a source free 3-D multilayered concentric cylindrical (MCC) model are presented. The fields were expanded in the modes of the MCC. Cost functions which specify mathematically the relative weight assigned to differences between an SAR distribution and a desired SAR distribution were defined. The coefficients of the modes, which minimize the cost function, were obtained using gradient search optimization methods. The optimized SAR distributions shown were computed using three different cost functions and two different radial locations for the center of the region where the desired SAR is largest. A five-layered model, including the outer water layer for cooling and improved matching with the source, was used. The frequency was 70 MHz. The current and charge distributions computed on a perfectly conducting cylindrical surface just outside the model are also shown. The surface current and charge distributions depend strongly on the relative importance of the cost for acute heat and systemic heat. A technique is developed for generating a new set of basis functions for reducing the number of unknowns to be optimized. We suggest that the approach shown could be useful in designing hyperthermia applicators.