Evaluation of a Dynamic Collimation System to Improve IMPT Dose Distributions and Maintain Treatment Efficiency

Background and objectives: Previous dynamic collimator system (DCS) developments included: (1) hardware construction and commissioning, (2) an accurate dose calculation algorithm, (3) a quality assurance approach, and (4) development of optimization tools for treatment planning. Clinical DCS implementation necessitates efficient treatment plan delivery and fully integrated tools. In this work, a novel algorithm for minimizing treatment time was developed with the goal of reducing the DCS time increase, relative to conventional pencil beam scanning, to one minute or less per beam. In this extensive end-to-end evaluation, treatment plans generated with a modified U.S. Food and Drug Administration (FDA)-cleared treatment planning system were delivered on an Ion Beam Applications (IBA) Proteus Plus proton therapy system, with and without a DCS, to evaluate delivery times and dosimetric accuracy for a relatively large patient dataset, providing evidence of the clinical potential of the approach. Methods: Ten previously treated brain patients were replanned, consisting of both deep-seated central and superficial targets, the latter of which required an external 4 cm polyethylene range shifter. DCS treatments were optimized using a maximum conformity planning technique exploiting per-spot collimator capabilities. An optimization algorithm was incorporated to minimize treatment delivery time by determining the optimal sequencing of spot positions and collimator settings. Plan quality was quantified using conformity and dose-volume histogram (DVH)-based metrics while delivery accuracy was validated through measurements using both patient-specific quality assurance (PSQA) and log file analysis at the Miami Cancer Institute (MCI). Results: The DCS reduced the dose gradient index on average by 26.4% (17.7-37.1%) and the mean dose to the adjacent healthy tissue (within 10 mm of the target) by 19.3% (16.3-26.2%). The average reduction to the mean and maximum dose to the involved optic nerves was 50% (25.7-80.7%) and 18.7%, respectively, and the mean and D_2cc dose to the involved brainstem was reduced by 63.9% (31.5-96.4%) and 60.4% (10.8-99.8%), respectively. PSQA pass rates among DCS-collimated and baseline uncollimated treatments were 99.7% and 99.2%, respectively. DCS treatment fields were delivered within an average of 49 s (32-61 s) from their uncollimated intensity modulated proton therapy (IMPT) counterparts. Average spot position errors were -0.05 ± 0.2 mm and 0.04 ± 0.2 mm for the x- and y-position, respectively. The maximum error in magnitude for collimator positioning was 0.2 mm or less. Conclusions: DCS collimated IMPT treatments can provide significant dosimetric improvements over uncollimated treatments. These highly collimated treatments can be delivered with sufficient accuracy for clinical use while incurring an additional time penalty of around one minute or less per field compared to uncollimated treatments.