Session Item

Pencil-beam scanned proton arcs (PBS arcs) have the potential to increase clinical benefit compared to conventional intensity modulated proton therapy (IMPT). To translate this new treatment technique from research to clinical practice, developments are needed on the treatment delivery systems. However, advances within treatment planning are equally important to enable optimization and delivery of PBS arc plans in a reasonable amount of time.

We propose a new method for Early Layer and Spot Assignment (ELSA) prior to spot weight optimization. The method analyzes the target from all directions simultaneously and assigns spots to a single energy layer per discretized direction. This substantially reduces the number of spot dose computations at the start of optimization and the number of variables in the subsequent optimization iterations compared to previous methods for PBS arc planning. To shorten the delivery time, ELSA sorts sequences of energy layers with respect to descending energy and applies a penalty on the number of upward energy jumps. We employed the new method for three prostate cancer patients in combination with robust optimization (21 scenarios with 3 mm setup and 3% range error) and Monte Carlo spot dose computation. The prescribed dose to the CTV was 77 Gy in 35 fractions. For each of the patients, three plans were constructed: 2-beam IMPT (2IMPT), 1-beam PBS arc (1Arc) and 2-beam PBS arc (2Arc) combining a CW arc with a CCW arc. Both PBS arc plans were discretized into approximately 180 gantry angles (2° spacing for 1Arc; 4° spacing for 2Arc). The resulting plans are assessed with respect to plan quality, optimization time and estimated delivery time.

Table 1 summarizes the results with respect to optimization and delivery time metrics, as well as clinical goal fulfillment. Both types of PBS arc plans reduce the integral dose to the patient. However, the 1Arc plan fails to fulfill the clinical goals for the target, especially in terms of robust target coverage (evaluated for the voxelwise minimum dose distribution over 42 scenarios). The 2Arc plans on the other hand outperform the 2IMPT plans in the nominal case and show similar robust target coverage as 2IMPT. The optimization time, including spot selection and spot dose computation, is longest for the 2Arc plans, but is below 6 minutes in all cases. The number of upward energy jumps are low for the PBS arc plans and the maximum delivery time is estimated to be just above 5 minutes. The trend for the energy layers and the relative segment weights for the PBS arc plans for patient 1 is shown in Figure 1.

We have developed a method for fast robust optimization of PBS arc plans. When utilizing energy layer sequencing with a penalty on the number of upward jumps, it is more favorable to deliver the arc over two revolutions. The two-beam arc approach results in high-quality plans, which can be optimized and delivered in a few minutes.