SU-F-T-359: Incorporating Dose Volume Histogram Prediction Into Auto-Planning for Volumetric-Modulated Arc Therapy in Rectal Cancer

Authors


Abstract

Purpose:

To incorporate dose volume histogram (DVH) prediction into Auto-Planning for volumetric-modulated arc therapy (VMAT) treatment planning and investigate the benefit of this new technique for rectal cancer.

Methods:

Ninety clinically accepted VMAT plans for patients with rectal cancer were selected and trained in the RapidPlan for DVH prediction. Both internal and external validations were performed before implementing the prediction model. A new VMAT planning method (hybrid_VMAT) was created with combining the DVH prediction and Auto-Planning. For each new patient, the DVH will be predicted and individual DVH constrains will be obtained and were exported as the original optimization parameters to the Auto-Planning (Pinnacle3 treatment planning system, v9.10) for planning. A total of 20 rectal cancer patients previously treated with manual VMAT (manual_VMAT) plans were replanned using this new method. Dosimetric comparisons were performed between manual VMAT and new method plans.

Results:

Hybrid-VMAT shows similar PTV coverage to manual_VMAT in D2%, D98% and HI (p>0.05) and superior coverage in CI (p=0.000). For the bladder, the means of V40 and mean dose are 36.0% and 35.6Gy for hybrid_VMAT and 42% and 38.0Gy for the manual_VMAT. For the left (right) femur, the means of V30 and mean dose are 10.6% (11.6%) and 17.9Gy (19.2Gy) for the hybrid_VMAT and 25.6% (24.1%) and 27.3Gy (26.2Gy) for the manual_VMAT. The hybrid_VMAT has significantly improved the organs at risk sparing.

Conclusion:

The integration of DVH prediction and Auto-Planning significantly improve the VMAT plan quality in the rectal cancer radiotherapy. Our results show the benefit of the new method and will be further investigated in other tumor sites.

Ancillary