2045 - A Novel Spot-Scanning Proton Arc(SPArc) Optimization Algorithm for Single Energy Extraction(SEE) Synchrotron-Accelerator-Based Proton Therapy System (PTS): Compelling Initial Results

Purpose/Objective(s): SPArc has drawn significant interest from the particle therapy community. However, all the existing SPArc optimization algorithms are only designed for cyclotron accelerators (PTS_-cyclotron) instead of synchrotron-accelerator-based Proton Therapy Systems with the Single Energy Extraction (SEE) technique (PTS_{-synchrotron-SEE}). The PTS_{-synchrotron-SEE} delivers an assortment of proton particles through each cycle. The remaining proton particles of that specific energy layer will be discarded if not used, which results in a significantly prolonged treatment delivery time. This study aims to develop the first SPArc optimization algorithm based on Dynamic Programming (SPArc_-DP), to improve the treatment delivery efficiency for PTS_{-synchrotron-SEE.}

Materials/Methods: Dynamic Programming, initially designed for combinational optimizations in fields of bioinformatics, finance and scheduling, was introduced to optimize the energy layer and MU distribution based on the features from the PTS_{-synchrotron-SEE}. Beginning from a plan generated via the original SPArc algorithm (SPArc_-original), based on the maximum charges per extraction, it iteratively merges the adjacent energy layers into the same energy layer while ensuring the original plan quality. Thus, it reduces the cycling from the PTS_{-synchrotron-SEE}. Five representative disease sites were selected for testing, including a base of skull chordoma, bilateral HN, prostate, lung, and liver cancers. The SPArc_-original plans were used as benchmarks. Dosimetric parameters, including target-dose conformity index (CI), target coverage, and sparing of OARs, were evaluated, along with the total number of cycles, utilization rate of each spill, as well as total treatment delivery time were simulated and compared between SPArc_-original and SPArc_-DP using a dynamic arc system controller.

Results: With a similar plan quality, the SPArc_-DP plans reduced the number of acceleration cycles by an average of 118.80 ± 72.15, or 237.60 ± 144.31s reduction in the total cycling time compared to SPArc_-original plans. These improvements effectively saved total treatment delivery time by 239.67s ± 218.15s (relatively 43% ± 13%) compared to SPArc_-original plans.

Conclusion: The study documents the first SPArc optimization algorithm for the PTS_{-synchrotron-SEE}. With a similar plan quality, the treatment delivery time has been shown to be significantly reduced for various disease subsites, which substantiates the value of the SPArc technique in the practical clinical practice of more efficient patient throughput for proton centers with synchrotron accelerator.