Conformal Dose Modulator for Proton Beam Therapy Part 1: A Simulation Study

• Published in: Frontiers in oncology 2022-04, Vol.12, p.840469
• Main Authors: Zhang, Zhexuan, Zhao, Weizhao, Butkus, Michael, Wu, Xiaodong
• Format: Article
• Language: English
• Subjects: 3D conformal modulator; compensator; Oncology; particle therapy; proton therapy; ridge filter
• ISSN: 2234-943X; 2234-943X
• DOI: 10.3389/fonc.2022.840469

Conformal dose deliveries in proton therapy utilize either a passive scattering system with a modulator or a pencil beam scanning (PBS) system. Efforts have been made to achieve conformal dose delivery by scanning a single energy layer of pencil beams through a 3D conformal modulator (3DCM), which combines a spread-out Bragg peak (SOBP) modulator consisting of a micro-pyramid array and a range compensator. The current published approach of designing such 3DCM relies on forward calculation methods to determine the geometry of the modulator. This study presents an alternative designing algorithm that inversely generates the geometry of a 3DCM paired with a corresponding fluence map, customized to patient-specific clinical indications. Critical spacing governing the size and separation between neighboring micro-pyramids was first determined, under which the dose homogeneity at desired depths could be achieved. We designed an adaptive ring optimization method using a modified gradient descent algorithm to inversely calculate the geometry of the 3DCM. This method includes several stages that progressively optimize both target coverage and dose conformity. The output contains the geometry of the 3DCM and its corresponding proton fluence map. Monte Carlo (MC) simulation was used to validate the results. The critical size and spacing of Lucite pyramids was determined to be 0.5 cm for a 184-MeV pristine proton beam. Using MATLAB (R2020a), the inverse designing algorithm generated an optimized 3DCM geometry and a fluence distribution achieving 100% target coverage with the 90% isodose surface and a corresponding conformity index of 1.057 on a spherical target. The resulting geometry was pruned to accommodate the MC simulation software and a currently accessible 3D printing service. The pruned geometry gave 95% target coverage by 90% isodose surface with a conformity index of 1.09 by ray-tracing dose computation. The MC simulation validated the 3DCM with 95% target coverage by 87% isodose surface and a conformity index of 1.12. We have demonstrated the feasibility of using a novel inverse optimization algorithm to generate 3DCM geometry and its corresponding proton beam fluence/intensity map, which could deliver highly conformal dose distribution with pencil beam scanning system using a single energy layer.