Solving the volumetric modulated arc therapy (VMAT) problem using a sequential convex programming method.

Area of Science:

• Medical Physics
• Radiation Oncology
• Computational Optimization

Background:

• Volumetric modulated arc therapy (VMAT) optimization is a complex, non-convex problem.
• Existing methods for VMAT face significant computational challenges compared to fixed-field intensity-modulated radiotherapy (IMRT).

Purpose of the Study:

• To develop and evaluate an efficient sequential convex programming (SCP) algorithm for VMAT optimization.
• To compare the performance of the SCP algorithm against ideal IMRT plans and exact solutions for VMAT.

Main Methods:

• A sequential convex programming algorithm is proposed, solving a series of convex optimization problems.
• The algorithm optimizes aperture weights and leaf positions using beam's eye view and constrained search spaces.
• It incorporates both global (large step-size) and local (small step-size) search strategies for improved accuracy and efficiency.

Main Results:

• The SCP algorithm generated VMAT plans comparable to ideal 72-beam IMRT plans in an average of 14 iterations and 36 minutes.
• For a down-sampled prostate case, the algorithm produced results equivalent to the exact non-convex mixed-integer optimization solution, but 12 times faster.
• The algorithm demonstrated scalability and applicability to real clinical cases.

Conclusions:

• The proposed SCP algorithm offers an efficient and effective solution for the challenging VMAT optimization problem.
• This method provides a practical approach for generating high-quality VMAT plans, outperforming exact methods in speed and scalability.
• The algorithm holds promise for clinical implementation in radiation oncology.