Accurate 3D-dose-based generation of MLC segments for robotic radiotherapy

Area of Science:

• Medical Physics
• Radiation Oncology
• Computational Biology

Background:

• Accurate radiotherapy dose modeling is crucial for treatment planning, accounting for factors like radiation scatter and multi-leaf collimator (MLC) characteristics.
• Fluence Map Optimization (FMO) generates intermediate plans using pencil beams, which are then converted into deliverable plans with MLC segments.

Purpose of the Study:

• To investigate novel approaches for segmenting FMO plans in robotic radiotherapy using a clinical dose engine (CDE).
• To develop and evaluate new methods for integrating CDE-calculated segment doses with pencil beams during the segmentation process.

Main Methods:

• Three versions of a segmentation algorithm were developed, varying in their integration of the CDE.
• New methods were created to combine CDE-derived segment doses with pencil beams for selecting new segments.
• Segmented plans were generated and compared against FMO plans for 20 prostate and 12 liver cancer patients.

Main Results:

• All three segmentation algorithm versions successfully mimicked FMO dose distributions.
• Segmentation with a fully integrated CDE yielded superior plan quality, with fewer monitor units and segments.
• The improved plan quality came at the expense of increased calculation time.

Conclusions:

• The developed segmentation algorithms effectively translate FMO plans into deliverable treatments.
• Full integration of a CDE in the segmentation phase offers the best trade-off for plan quality and efficiency in robotic radiotherapy.
• Further optimization may be needed to balance plan quality with computational time.