3D numerical simulations on GPUs of hyperthermia with nanoparticles by a nonlinear bioheat model

This paper deals with the numerical modeling of hyperthermia treatments by magnetic nanoparticles considering a 3D nonlinear Pennes’ bioheat transfer model with a temperature-dependent blood perfusion in order to yield more accurate results. The tissue is modeled by considering skin, fat and muscle layers in addition to the tumor. The FDM in a heterogeneous medium is employed and the resulting system of nonlinear equations in the time domain is solved by a predictor–multicorrector algorithm. Since the execution of the three-dimensional model requires a large amount of time, CUDA is used to speedup it. Experimental results showed that the parallelization with CUDA was very effective in improving performance, yielding gains up to 242242 times when compared to the sequential execution time.