Application of control theory to the hyperthermia problem.

Abstract

The objective of a hyperthermia cancer treatment is to heat the tumor tissue to a therapeutic level while limiting the detrimental effects experienced by the surrounding normal tissue. To achieve an optimal treatment requires knowledge of the resulting temperature response and an understanding of the complex interaction between the thermal response, the applied power, and the blood flow in the target tissue region. This dissertation considers model reduction to overcome the large dimensions associated with thermal modelling, extended Kalman filtering to estimate both the unmeasured temperature states and the unknown blood perfusion magnitudes, optimization of the applied power to achieve the best thermal response, and optimal servomechanism control to attain the desired regulated output tracking. A controller methodology that combines thermal estimation, applied power optimization, and optimal servomechanism control with a simple expert system shell is examined. This controller methodology is analyzed for a simulated scanned focussed ultrasound system (SFUS) based upon the bioheat transfer equation (BHTE) model of the thermal response in the target region. The results of the presented studies illustrate the following important points. First, open-loop reduced-order models based on the balanced transformation provide drastic model reduction for controller design purposes. Second, the success of thermal estimation depends on the number and the location of the thermal sensors, and the accuracy of the modelled blood perfusion profile. Third, multiple modelling in estimation provides an alternate technique for overcoming model mismatch associated with the modelling of the blood perfusion pattern. Fourth, the choice of the set points for the optimal servomechanism controller play a crucial role in the resulting tissue temperatures. Fifth, the scan parameter sets that result in optimal SFUS power profiles need to be changed on-line during a treatment as the blood perfusion magnitude and pattern are estimated. Finally, to fully automate a hyperthermia treatment requires that the expertise of the clinician be incorporated into the controller design. Hierarchical control provides a means of incorporating the expert system shell at the higher levels of the controller, while maintaining optimal servomechanism control at the lower levels.