Identification of tissue and therapy parameters from surface temperature

Abstract

The temperature distribution of human body tissues, when electromagnetic waves are applied, is studied with the help of the bioheat equation, which is characterised by several thermal parameters, such as thermal conductivity, perfusion frequency or metabolic heat. There are also electromagnetic parameters, such as electrical conductivity and dielectric constant. Besides, therapy parameters, such as the applied power should be considered. If the values of all these parameters are known, the time evolution of the temperature can be determined. Some of these values can be found in specialised databases, and others are unknown but approximate values can be obtained by reasonable estimations. The outcome of simulations depends heavily on the parameter values. Surface temperature helps in providing better estimates of the parameter values. Once these parameters are identified, medical analyses can be performed to assess the dosimetry of the radiation so that it does not damage tissues, for example. The surface temperature is obtained from a sequence of thermographic images. Based on these experimental data, an algorithm is applied to find the values of the needed parameters. The model is simulated iteratively, adjusting the parameters at each step, reducing the approximation error between the simulation and the data. This is an optimization problem that belongs to the realm of inverse problems. It can be solved using techniques based on the gradient concept, however, this problem can be ill-conditioned, so probabilistic or evolutionary algorithms are also used. In this paper, the simulation is made using a method based on Legendre wavelets. It is proposed that the subsequent optimization is made using an evolutionary algorithm, that has shown great robustness in the problems where it has been applied. As far as known, it has never been applied to the bioheat equation. This is the aim of this work.