Knowledge-based DVH estimation and optimization for breast VMAT plans with and without avoidance sectors

Area of Science:

• Medical Physics
• Radiation Oncology

Background:

• Knowledge-based treatment planning systems, like RapidPlan, are used for VMAT in breast cancer.
• Avoidance sectors (AS) are a feature imposed during optimization to minimize dose to organs at risk.

Purpose of the Study:

• To analyze RapidPlan models for DVH estimation in breast cancer VMAT plans with avoidance sectors.
• To evaluate the impact of avoidance sectors on DVH prediction accuracy.

Main Methods:

• Two VMAT plan types (PartArc and AvoidArc) were generated for 20 left breast cancer patients.
• Two RapidPlan models were created from these plans and evaluated in a closed loop.
• Model performance was assessed with and without avoidance sectors during optimization.

Main Results:

• The PartArc model accurately estimated DVHs. The AvoidArc model showed discrepancies when avoidance sectors were not included in training.
• Inaccurate DVH estimations for contralateral structures were observed when the AvoidArc model was trained without AS.
• When AS were included in training and optimization, the AvoidArc model provided comparable DVH estimations.

Conclusions:

• RapidPlan models can accurately estimate organ at risk DVHs for VMAT breast cancer planning.
• Proper plan selection and usage for model training are essential, particularly when incorporating avoidance sectors.