A Core Body Temperature Retrieval Method for Microwave Radiometry when Tissue Permittivity is Unknown.

This paper presents a novel method for core temperature retrieval using microwave radiometry when complex permittivity and heat transfer parameters of the tissue layers of the human subject are unknown. Previous works present methods for core temperature retrieval, but these methods do not account for population variation in the relevant electromagnetic and thermal parameters, which can increase measurement error beyond the clinically acceptable limit of 0.5°C. Pennes' bioheat model of a six-tissue-layer human head model combined with a coherent electromagnetic model simulate experimental data. To retrieve core temperature, nonlinear least squares optimization is then used to minimize the difference between the simulated experimental data and an exponential model for physical temperature and the coherent electromagnetic model. By using 20 frequencies spanning from 1-5 GHz, core temperature is retrieved while accounting for population variation in the permittivity and thermal parameters. A Monte Carlo simulation in which the thermal parameters and permittivity vary according to literature-derived, population-representative distributions and the core body temperature varies from 18-46°C is used to assess the utility of the retrieval method. Different antenna patterns are tested to explore the effect on retrieval accuracy. The retrieval method has a retrieval error of <0.1°C when only the thermal parameters are unknown and a retrieval error of <0.5°C when the thermal parameters and permittivity are unknown, which is within the clinically acceptable error range of 0.5°C. These results help progress the field of medical microwave radiometry toward being a clinically viable noninvasive measurement that is accurate when measuring all patients.