FRoG dose computation meets Monte Carlo accuracy for proton therapy dose calculation in lung

Area of Science:

• Medical Physics
• Radiation Oncology
• Computational Biology

Background:

• Proton therapy, particularly spot-scanning intensity-modulated proton therapy (IMPT), offers precise dose delivery.
• Accurate dose calculation is critical for effective and safe IMPT, especially in complex geometries like the lung.
• Existing analytical models may have limitations in capturing the full physics of proton interactions.

Purpose of the Study:

• To benchmark and evaluate novel analytical GPU-accelerated and CPU-based Monte Carlo (MC) dose-engines for IMPT.
• To assess the clinical viability of these dose-engines for lung cancer treatment.
• To compare the performance of FRoG, RayStation-MC (RS-MC), and general-purpose MC (gp-MC) dose calculation systems.

Main Methods:

• Nine patient cases from clinical experience and a public database were used for in-silico study.
• Plans were optimized using 2 orthogonal beams in RayStation v.8.
• Forward calculations were performed with FRoG (analytical PBA), RS-MC (CPU), and gp-MC. Dosimetric benchmarks were obtained using a lung-like phantom and ionization chambers.

Main Results:

• FRoG and RS-MC showed average GTV dose differences of 1.8% and 2.3% compared to gp-MC.
• FRoG and RS-MC achieved local γ-passing rates of approximately 96% and 93%, respectively.
• Phantom measurements confirmed FRoG's high accuracy with deviations under 0.1%.

Conclusions:

• GPU-accelerated analytical PBA, MC-TPS, and gp-MC dose calculations are within clinical tolerances.
• FRoG predictions closely matched gp-MC and experimental data for thoracic proton dose calculations.
• FRoG offers a viable solution for clinical treatment planning or as an auxiliary dose-engine for lung cancer patients.