Scalable, memory-efficient robust proton therapy optimization through beamlet-free treatment planning.

Area of Science:

• Medical Physics
• Radiation Oncology
• Computational Biology

Background:

• Proton therapy optimization must account for treatment uncertainties.
• Current robust optimizers use worst-case scenarios, increasing computation time and memory needs due to beamlet matrix calculations.

Purpose of the Study:

• Investigate a beamlet-free robust optimization strategy.
• Reduce computational requirements by performing optimization and scenario evaluation during Monte Carlo dose calculation, omitting the beamlet matrix.

Main Methods:

• Compared three robust optimization methods: beamlet-based worst-case, beamlet-based expected-value, and beamlet-free expected-value.
• Evaluated methods on six patient cases for plan quality and computational burden.

Main Results:

• Achieved comparable plan quality across all methods.
• Beamlet-free robust optimization reduced memory usage by 87-92% and computation time by 51-69% compared to beamlet-based methods.

Conclusions:

• Beamlet-free robust optimization maintains plan quality while lowering computational demands.
• This method supports more complex optimization conditions, potentially advancing novel proton therapy techniques.