Novel Optimization Methods and Treatment Planning System for Clinically-Deliverable Truly-Hybrid Proton-Photon Radiotherapy

项目摘要 放射治疗(RT)的目的是将杀瘤剂量输送到临床靶区(CTV)，同时在 风险(OAR)，对此质子和光子束自然是相辅相成的：质子 通常更有利于桨的备用，而光子对CTV覆盖的交付不确定性更强大。 质子-光子混合RT有着悠久的历史。然而，由于它分别产生质子和光子计划 在规划阶段没有充分利用质子-光子联合优化，目前的混合RT是伪随机的。 混合型，在计划质量、治疗地点和广泛适用性方面非常有限。跨越的关键是从 伪杂化到真杂化RT是一种新的协同质子-光子联合优化方法 互补的质子和光子束。假设是真正的杂化RT通过适当的联合质子- 在CTV覆盖方面，光子优化将比质子或纯光子RT更有利 稳健性和划桨节省最佳性。 真正混合放射治疗对患者的广泛适用性：(A)临床适用性：与伪混合放射治疗不同， 由于计划质量和治疗地点的限制，真正的杂交RT可能成为普通癌症RT的新范例， 由于其优越的计划质量，因此与仅质子或仅光子RT相比，具有潜在的临床结果。 (B)临床工作流程：我们的真正混合计划可以单独安全地在现有质子和 光子机器，这项努力设想患者在像我们这样的综合癌症中心接受治疗， 质子和光子设备，在一名医生的指导下，使用共享固定化 设备、模拟程序和结构设置，以及综合治疗计划和实施系统。 (C)患者覆盖面：真正的混合放射治疗可以通过现有的 美国的基础设施，因为(1)大多数有质子中心的医院也有光子中心；(2)76%的癌症 患者生活在质子中心正在运作的州，而85%的患者在100英里(2小时车程)内 到这些质子中心的距离；(3)癌症患者更愿意旅行寻求先进的治疗方案。 建议的工作：受前所未有的计划质量和真正混合RT的广泛适用性的启发，通过我们的 联合质子-光子优化方法，下一步是通过临床试验对假设进行前瞻性检验。 然而，将真正的杂交RT从研究推进到临床的一个缺失的先决条件是治疗计划 可生成临床可交付的混合计划的系统(TPS)。为了满足这一迫切需求，这一努力将 为临床可交付的真正杂交RT开发新的优化方法和TPS，这是根本的一步 为验证这一假说而进行的前瞻性临床试验。 目的1：临床可交付的真杂交RT的优化方法和TPS。 目标2：精确高效的MCO真混合规划的优化方法。 目标3：基于深度学习的高效真混合规划优化方法。

Project Summary Radiotherapy (RT) aims to deliver tumoricidal dose to clinical target volume (CTV) while sparing organs at risk (OAR), for which proton and photon beams are naturally complementary to each other: protons are generally better for OAR sparing, while photons are more robust to delivery uncertainties for CTV coverage. The hybrid proton-photon RT has a long history. However, as it generates proton and photon plans separately without fully utilizing joint proton-photon optimization during the planning stage, current hybrid RT is pseudo- hybrid and very limited in plan quality, treatment sites, and broad applicability. The key to leapfrog from pseudo-hybrid to truly-hybrid RT is new joint proton-photon optimization method that synergizes complementary proton and photon beams. The hypothesis is that truly-hybrid RT via appropriate joint proton- photon optimization will be more favorable than proton or photon-only RT, in terms of CTV coverage robustness and OAR sparing optimality. Broad applicability of truly-hybrid RT to patients: (A) Clinical applicability: unlike pseudo-hybrid RT that is limited in plan quality and treatment sites, truly-hybrid RT may become a new paradigm for general cancer RT, owing to its superior plan quality and thus potentially clinical outcomes to proton-only or photon-only RT. (B) Clinical workflow: our truly-hybrid plans can be individually and safely delivered on existing proton and photon machines, and this effort envisions patients being treated in an integrated cancer center like ours with both proton and photon equipment, under the direction of a single physician, using shared immobilization devices, simulation procedure and structure set, and integrated treatment planning and delivery system. (C) Patient coverage: truly-hybrid RT can be made broadly available to many cancer patients through existing infrastructures in US, since (1) most hospitals with proton centers also have photon centers; (2) 76% of cancer patients live in the states with operational proton centers, while 85% are within 100-mile (2-hour-driving) distances to these proton centers; (3) cancer patients are more willing to travel for advanced treatment options. Proposed effort: Inspired by unprecedented plan quality and broad applicability of truly-hybrid RT via our joint proton-photon optimization method, the next step is to test the hypothesis prospectively via clinical trials. However, a missing prerequisite to advance truly-hybrid RT from research to clinic is a treatment planning system (TPS) that can generate clinically-deliverable hybrid plans. To meet this urgent need, this effort will develop novel optimization methods and TPS for clinically-deliverable truly-hybrid RT, which is a radical step towards prospective clinical trials for testing the hypothesis. Aim 1: Optimization methods and TPS for clinically-deliverable truly-hybrid RT. Aim 2: Optimization methods for accurate and efficient MCO truly-hybrid planning. Aim 3: Deep learning based optimization methods for efficient truly-hybrid planning.