2440 - Mid-Range Arc Therapy—Enhancing Dosimetric Quality, Robustness and Efficiency for Proton Treatment with Simple Planning

Q. Wang¹, M. Chen¹, M. Kazemimoghadam¹, X. Gu², and W. Lu¹; ¹Medical Artificial Intelligence and Automation (MAIA) Lab, Department of Radiation Oncology, UT Southwestern Medical Center, Dallas, TX,, Dallas, TX, ²Stanford University Department of Radiation Oncology, Palo Alto, CA

Purpose/Objective(s): Proton arc therapy (PAT) using spot-scanning shows promise in clinical benefits, but faces challenges in treatment robustness and efficiency as well as control the optimization complexity within reasonable order of magnitude. To address these challenges, we present a novel mid-range proton arc therapy (MRPAT) strategy with monoenergetic beam per control point and a low-complexity global optimization framework. Materials/

Methods: In mid-range beam selection, our principal aim is that Bragg peaks can sweep across the central part of the target after full arc delivery. Thus, the sole mid-range energy layer for each control point is defined as the beam range through the target isocenter, measured by the water equivalent path length (WEPL). The advantages of using mid-range beam are three-fold. First, positioning mid-range spots near the target center establishes a buffer zone, reducing uncertainties such as proton beam range, respiratory motion, and setup errors. Second, the delivered dose primarily comes from Bragg peak, enhancing linear energy transfer (LET) for greater relative biological effectiveness (RBE). Last is efficient delivery, as the WEPL varies smoothly with the beam angle, energy layer switching would take little time and be performed while gantry rotating. For spot intensity optimization, we adopted the global optimization framework previously presented by our team. The spot dose is modeled by kernel decomposition (KD), and calculation is speeded up by Fast Fourier Transform. In contrast to conventional fluence map optimization that needs to pre-calculate and store a huge dose deposition matrix, the KD model only requires to store a shift-invariant kernel obtained from the planning tool method, which greatly reduces both spatial and temporal complexities. The feasibility study of the MRPAT framework was performed on a two-dimensional cylindrical homogeneous phantom with a centrally located target surrounded by four organs at risk (OARs). The MRPAT plan was compared with a single-field (field-1) plan and a conventional PAT (full-arc) plan both with full-energy.
Results: The optimization results are summarized in the table. The field-1 plan received more dose in the entrance path and OARs compared with two arc plans. The MRPAT plan achieved similar target coverage and improved OAR sparing compared with full-arc PAT plan, but MRPAT utilized less energy layer and spots. Furthermore, it is important to note that MRPAT plan saves 10-fold computation time compared to the full-arc PAT plan.
Conclusion: The novel MRPAT framework demonstrated its ability to generate plan quality comparable to full-energy PAT delivery within less optimization time. Key advantages of MRPAT include offering the potential to maximize LET for enhanced RBE and establishing robust plans by leveraging mid-range spots. Abstract 2440 – Table 1

Plan	# Energy	# Spots	Intensity	PTV D2 (Gy)	PTV D98 (Gy)	OAR D2 (Gy)	HI	CalcTime (s)
Field-1	21	330	648	120	96	85	0.24	24
Full-arc	26	24626	670	121	98	22	0.23	7426
MRPAT	1	1756	795	120	99	14	0.21	690