CONTROLLING MR-GUIDED HYPERTHERMIA THERAPY FOR CANCER TREATMENT WITH KALMAN FILTER AND INCREASED SPATIAL RESOLUTION

This work addresses the optimal control of magnetic resonance-guided hyperthermia therapy for cancer treatment in biological tissues. The proposed methodology consists of a synergy between noninvasive heating and temperature measurement techniques, the former being controlled in terms of the deviation of the measured temperature from a designed distribution. The latter is assumed available via magnetic resonance thermometry, whose data are filtered through a state estimation technique to reconstruct the temperature data within the tissue, using the Pennes bioheat transfer equation in the evolution model. This step, performed in terms of the classical Kalman filter (KF), ensures that the spatial resolution from the magnetic resonance thermometry can be significantly increased, which proves to be paramount, especially when dealing with ablation. The evolution and observation matrices in the KF were automatically assembled using modified versions of the solver laplacianFoam and the subroutine mapFields, respectively, both from the Open-FOAM software. The heating source is controlled by a proportional-integral-derivative (PID) algorithm in order to ensure that the set point temperature is achieved and maintained. A numerical experiment was conducted, in which synthetic measurements were considered available. The results showed that the proposed framework successfully controlled the temperature, even showing robustness to random disturbances in the current estimate. This methodology is also easily extendable to realistic problems, with three-dimensional anatomically coherent geometries. Finally, it has a significant potential for high-performance computing improvements, due to the matrix nature of the KF and the spar-sity of its matrices.