Feasibility of estimating the temperature distribution in a tumor heated by a waveguide applicator.

The feasibility of using a 2-dimensional (2D) modeling approach for retrospectively describing complete temperature distributions in the midplane of a tumor during a clinical hyperthermia treatment was tested. An experimental treatment, using a 915-MHz waveguide applicator to heat a large melanoma in a dog, was modeled. Detailed measurements of temperatures were made during the treatment. The steady-state blood flow distribution at the midplane was imaged by positron emission tomography (PET), and these data were used to prescribe the modeled perfusion pattern. A 2D finite element method (FEM) was used to approximate the solution to Maxwell's Equations to obtain the specific absorption rate (SAR) distribution. The blood-flow estimates, assumed material properties, SAR distribution, and temperature boundary conditions were then used with the same mesh in a second FEM program to obtain a solution to the bioheat transfer equation. This latter routine was embedded in a state-and-parameter-estimation program that systematically varied selected parameters until the differences between computed and measured temperatures were minimized. Optimizations were performed independently for three subsets of the measured temperature data to assess the sensitivity of the predicted temperature field to the number of measurements. The calculated temperature distributions that resulted were similar to each other, and the predicted temperatures at the sensor points excluded from these optimizations were in reasonable agreement with the measurements. However, lack of unique blood flow values following optimization indicates that the methods of estimating blood flow will need to be improved or that there are problems with model mismatch. This work is a clinical case study of an evolving 2D system of thermal dosimetry which relies on both empirical and theoretical concepts. The methodology is being evaluated for its ability to generate prognostically significant descriptors of the treatment temperature field.