Integrating DVH criteria into a column generation algorithm for VMAT treatment planning.

Area of Science:

• Medical Physics
• Radiation Oncology
• Computational Optimization

Background:

• Volumetric-modulated arc therapy (VMAT) offers flexible radiation delivery but presents complex optimization challenges.
• Achieving clinically acceptable VMAT plans often requires extensive manual adjustments and trial-and-error.
• Prioritizing tissue dose constraints within VMAT optimization remains a significant hurdle.

Purpose of the Study:

• To develop an automated VMAT planning algorithm that integrates dose-volume histogram (DVH) criteria directly into the optimization process.
• To reduce reliance on manual intervention and iterative plan adjustments for VMAT.
• To enhance the efficiency and robustness of VMAT treatment planning.

Main Methods:

• Implementation of a direct aperture optimization algorithm based on column generation.
• Iterative adjustment of penalty function weights guided by DVH criteria during optimization.
• Evaluation of the algorithm's performance and plan quality using clinical prostate and head-and-neck cancer cases.
• Assessment of algorithm robustness through multiple random initial weight vector configurations.

Main Results:

• The automated method successfully generated clinically acceptable VMAT plans for all tested prostate cases, outperforming standard VMAT optimization.
• For a challenging head-and-neck case, the algorithm achieved 93.3% clinically acceptable plans, compared to none with standard VMAT.
• The proposed approach demonstrated significantly better plan quality, particularly for organs at risk, and reduced computational time compared to commercial software.
• The algorithm proved robust to variations in initial weight parameters.

Conclusions:

• The developed optimization algorithm effectively generates clinically acceptable VMAT plans automatically and robustly.
• The integrated DVH-based weight adjustment procedure streamlines the planning process and improves plan quality.
• This automated approach minimizes the need for manual post-optimization adjustments, saving time and effort in VMAT planning.