Using OVH and IMRT Plan Data to Automate VMAT Planning: A Head-And-Neck Study
b wu1*, D Pang2, g sanguineti3, R Taylor4, T McNutt5, (1)(2)Georgetown University Hospital, (3)(4)(5)Johns Hopkins UniversitySU-C-211-3 Sunday 1:30:00 PM - 2:15:00 PM Room: 211
Purpose: Although VMAT is widely used, VMAT planning faces challenges similar to those of IMRT planning, particularly with respect to planning efficiency and plan quality. The study is to investigate whether VMAT planning efficiency and quality can be improved through an overlap volume histogram (OVH)-driven, automated planning application using an IMRT database.
Methods: Based on comparable dosimetric results between planner-generated head-and-neck VMAT and IMRT plans, an in-house developed OVH-driven, automated planning application containing a database of 182 prior clinical head-and-neck IMRT plans manually created by dosimetrists is built into Pinnacle SmartArc to automate VMAT planning on an Elekta Infinity Linac. The only required inputs to the application are the contours of the CTVs and OARs delineated on a patient's planning CT. All the planning tasks are fully automatically executed by the application without any user intervention. Double arcs, 2 degree sampling frequency of gantry, 180 s/arc maximum delivery time, 45 degree collimator angle and 0.25 cm/deg constrain leaf motion are used in VMAT planning. VMAT plans of 4 oropharynx, 4 nasopharynx and 4 larynx patients are fully automatically generated and compared with corresponding clinical IMRT plans in the database. Efficiency is evaluated by the number of optimization rounds required for a complete plan.
Results: All VMAT plans are automatically generated in two optimization rounds, while the average optimization rounds in clinical IMRT oropharynx, larynx and nasopharynx planning are 28, 36 and 67 respectively. In VMAT plans, significant dose reductions (p<0.05) to the cord+4mm (D0.1cc: 3.7 Gy), brainstem (D0.1cc: 4.9 Gy), brachial plexus (D0.1cc: 1.6 Gy), larynx (V(50 Gy): 5.3%) and inner ear (mean dose: 4.4 Gy) with a slight degradation in low-level PTV coverage (V95: 0.3%; p=0.25) are observed.
Conclusions: IMRT-data-driven VMAT planning is an efficient and effective way to automate VMAT planning.