A Novel IMRT Plan Optimization Algorithm for Physician-Driven Plan Tuning
M Zarepisheh*, N Li, L Cervino, K Moore, X Jia, S Jiang, Center for Advanced Radiotherapy Technologies, University of California, San Diego, La Jolla, CASU-E-T-588 Sunday 3:00PM - 6:00PM Room: Exhibit Hall
To greatly speed up the time-consuming treatment planning procedure by developing a new optimization algorithm allowing the physicians to interactively fine-tune the DVHs and iso-dose lines of an IMRT plan.
The conventional treatment planning procedure is a time-consuming and resource-demanding task that may need multiple iterations between the dosimetrists and clinicians after an initial plan is developed. We develop a new optimization algorithm to speed up this procedure by allowing the physician to interactively fine-tune the DVH and iso-dose lines on top of the initially optimized plan. After the physician modifies the DVH and iso-dose lines of the current plan towards a more desired one through an interactive graphical interface, the algorithm will adjust the voxel-dependent optimization parameters to guide the plan towards the modified one. To ease the fine-tuning procedure, the algorithm enables the physician to lock some DVH curves as well as to change the priorities of the organs. The algorithm explores a large Pareto surface by adjusting voxel-dependent parameters to find out a plan that is the closest to the physicians desired plan.
The algorithm was tested using a series of clinically realistic patient data and found to have desirable performance. It can adjust the voxel-dependent parameters and guide a plan towards the physician-driven fine-tuned plan. This algorithm has been implemented on GPU for high efficiency, and the updating procedure is near real time.
The conventional treatment planning procedure can be significantly improved in terms of efficiency and physician satisfactory by utilizing the new GPU-based algorithm allowing the physician to interactively fine-tune the plan in near real time.
Add this talk to vcal | ical | Contact Email: