Robotische Systeme und Regelungsstrategien für die Radiotherapie bewegter Tumore

  • Robotic systems and control strategies for radiotherapy of moving tumors

Arenbeck, Henry; Abel, Dirk (Thesis advisor); Corves, Burkhard (Thesis advisor)

Aachen : Shaker (2015)
Book, Dissertation / PhD Thesis

In: Berichte aus der Medizintechnik
Page(s)/Article-Nr.: XI, 256 S. : Ill., graph. Darst.

Zugl.: Aachen, Techn. Hochsch., Diss., 2015


Radiotherapy of moving tumors is currently undergoing a generation shift towards 4D techniques. These techniques allow a both volume and time conformal application of radiation dose and correspondingly require an intra fractional closed loop control of dose application. A 4D radiotherapy can be realized by positioning of the couch, which carries the patient under radiation, by a serial industrial robot with the aim of eliminating the inertial motion of the irradiated volume. This principle allows a high accuracy of motion compensation. Also, the large working space of a serial industrial robot enables efficient patient logistics and extended degrees of freedom of the radiation program.In this work, a novel 4D phantom, which generates high dimensional motion of the irradiated volume relative to its surrounding human equivalent environment, is presented. Also, prediction and control methods, which allow robot-based compensation of respiratory motion, are presented and compared based on both theory and experiments. The main desired features of a 4D phantom are: (1) Human equivalent structure, (2) human equivalent motion, (3) comprehensive measurement capabilities, (4) universal applicability and (5) efficient practical use. The 4D phantom presented in this work displays high performance with respect to each of these features. The 4D phantom is modularly structured, allows a sixdimensional physiological motion of the irradiated volume and employs a new kind of parallel robot, which allows a customization of the generated motion in three degrees of freedom. The 4D phantom was successfully employed in clinical practice for validation of various techniques of medical imaging and radiotherapy. Various prediction methods, which are based on adaptive linear filters, local regression, adaptive interval type two neuro fuzzy inference systems or models of the expected propagation of the predicted value, were implemented, benchmarked and compared considering the requirements of the 4D radiotherapy application. For this purpose, exemplary human motions and eleven benchmark criteria, which were evaluated based on thirty-eight different property-values, were employed. Finite impulse response filters, whose coefficients are continuously updated using the recursive least squares method, were identified as best suited.Robot based inertial compensation of respiratory motion can be realized based on: (1) Adaptive identification and compensation of robot latency, (2) combined feedforward and feedback control, (3) model based predictive control and (4) period robust repetitive control. The theoretical advantages and disadvantages of these control methods were complemented by an analysis of the practical results obtained with these methods. In that context, the deviations from a periodic characteristic of human respiratory motion were statistically quantified. The extent of these deviations rendered repetitive control variants unprofitable. Combined feedforward and feedback control and model based predictive control were identified as best suited. The robot of the 4D phantom, a selected prediction method, a selected control method, a serial industrial robot and an optical tracking system were combined to a representative substitute system of robot based patient couch positioning for compensation of respiratory motion. Feasibility and benefit of such system were demonstrated under real conditions.