Cardinality-constrained plan-quality and delivery-time optimization method for proton therapy.

Area of Science:

• Medical Physics
• Radiation Oncology

Background:

• Minimizing proton therapy plan delivery time is crucial for motion management, patient comfort, and treatment throughput.
• Plan delivery time is often limited by energy switching time, which is proportional to the number of energy layers (cardinality).

Purpose of the Study:

• To develop a novel optimization method for efficiently computing the Pareto surface between plan quality and energy layer cardinality.
• To enable planners to navigate the tradeoff between plan quality and efficiency and select an optimal plan.

Main Methods:

• Proposed a new Intensity Modulated Proton Therapy (IMPT) method called CARD.
• CARD explicitly incorporates minimization of energy layer cardinality using l1,0-norm regularization and automatically generates a sequence of plans with decreasing cardinality.
• Plan optimality and deliverability were enforced using dose-volume objectives and minimum monitor unit (MMU) constraints, with an iterative convex relaxation algorithm.

Main Results:

• CARD efficiently computed plans with reduced energy layers (e.g., 52% reduction for prostate cases) compared to a benchmark plan (P0) while maintaining acceptable plan quality.
• CARD outperformed a state-of-the-art method (MMSEL) in automatic plan generation and efficiency, requiring no manual parameter tuning.
• CARD achieved similar or better optimization objectives than P0 with fewer energy layers, demonstrating its effectiveness in balancing quality and efficiency.

Conclusions:

• Developed a novel optimization algorithm, CARD, for efficient and automatic computation of treatment plans with sequential energy layer reduction.
• CARD enables planners to effectively navigate the tradeoff between plan quality and energy layer cardinality.
• The method facilitates the selection of balanced treatment plans for proton therapy.