This Ph.d. project describes the development of a workflow for Monte Carlo Treatment Planning for clinical radiotherapy plans. The workflow may be utilized to perform an independent dose verification of treatment plans. Modern radiotherapy treatment delivery is often conducted by dynamically modulating the intensity of the field during the irradiation. The workflow described has the potential to fully model the dynamic delivery, including gantry rotation during irradiation, of modern radiotherapy. Three corner stones of Monte Carlo Treatment Planning are identified: Building, commissioning and validation of a Monte Carlo model of a medical linear accelerator (i), converting a CT scan of a patient to a Monte Carlo compliant phantom (ii) and translating the treatment plan parameters (including beam energy, angles of incidence, collimator settings etc) to a Monte Carlo input file (iii). A protocol for commissioning of a Monte Carlo model of a medical linear accelerator, ensuring agreement with measurements within 1% for a range of situations, is presented. The resulting Monte Carlo model was validated against measurements for a wider range of situations, including small field output factors, and agreement with measurements within 1–2% was found. Although the protocol was applied to a specific accelerator type it can be applied to any medical linear accelerator with similair design. A new algorithm for converting CT scan of a patient to a Monte Carlo compliant phantom is presented. It is more sophisticated than previous algorithms since it uses delineations of structures in order to include and/or exclude certain media in various anatomical regions. This method has the potential to reduce anatomically irrelevant media assignment. In house MATLAB scripts translating the treatment plan parameters to Monte Carlo input files were written. The scripts are tested and validated for modern treatment delivery including multi leaf collimator movement and gantry rotation during irradiation. Moreover, a workflow binding the elements together and thus enabling Monte Carlo Treatment Planning is presented. Comparison between dose distribution for clinical treatment plans generated by a commercial Treatment Planning System and by the implemented Monte Carlo Treatment Planning workflow were conducted. Good agreement was generally found, but for regions involving large density gradients differences of 6% were observed.
Main Research Area:
E. Andersen, Claus, F. Behrens, Claus, Helt-Hansen, Jakob