Deliverable Optimization Incorporating Monitor Unit Constraints for Intensity Modulated Proton Therapy Treatment Planning
X Li1*, W Cao2, X Wang1, Y Li1, X Zhu1, N Sahoo1, M Gillin1, G Lim2, X Zhang1, (1) The University of Texas MD Anderson Cancer Center, Houston, TX, (2) University of Houston, Houston, TXTH-A-213AB-11 Thursday 8:00:00 AM - 9:55:00 AM Room: 213AB
Purpose: Delivery constraints on monitor units (MUs) are not currently considered during optimization of intensity modulated proton therapy (IMPT) planning in our treatment planning system (TPS). This may result in differences between the delivered and ideally optimized dose distributions. The purpose of this study is to develop a deliverable optimization algorithm incorporating MU constraints (DOMU) in IMPT treatment planning and evaluate its clinical applicability.
Methods: A novel objective formula was incorporated with the limited-memory Broyden-Fletcher-Goldfarb-Shannon (L-BFGS) algorithm to directly optimize the deliverable IMPT plan with minimum MU constraint (0.005 MUs per spot). Treatment plans for a homogeneous phantom were generated for different spot spacings, using optimization without MU constraints and post-processing (OMPP) and DOMU. A treatment plan for a head-neck cancer patient prescribed to 63 Gy was also designed with DOMU using different spot spacings. For all plans, the dose-volume-histograms (DVHs) were analyzed to evaluate the benefits of the DOMU.
Results: For the regular shaped homogeneous phantom, only the plan generated by OMPP using 5mm spacing showed pronounced dose distortion compared to ideally optimized dose. Plan using 7mm spacing still produced accepted target coverage. For the patient study, plans generated by OMPP using both 5mm and 7mm spacing displayed significant dose distortion and unacceptable target coverage. The dose covering 99% of the target volume was 60.2 Gy and 61.7 Gy for 5mm and 7mm spot spacing plans generated by DOMU, an increase of 18.8 Gy and 18.4 Gy compared with plans generated by OMPP. The parotid sparing of DOMU generated plans in 5 mm spot spacing was significantly better than that in 7 mm spot spacing indicating the importance of using smaller spacing to improve the IMPT quality.
Conclusion: The DOMU algorithm, which incorporates the minimum MU constraints into the optimization, can reduce the dose distortion caused by post-processing process and achieve the optimum delivered dose.