One-click Automated 3D Treatment Planning for Radiopharmaceutical Therapy

项目摘要/摘要 预计到2025年，放射药物治疗(RPT)市场将超过Technitium-9900万市场， 重点是转移性前列腺癌、神经内分泌肿瘤和淋巴瘤。在井的推动下- 可以耐受的治疗和更少的副作用，专家估计将需要150个新的治疗中心 在美国，估计每年提供50,000-200,000个治疗周期。目前，所有的RPT治疗 尽管患者的生理和药代动力学独特，但仍需给予标准的治疗剂量。从… 根据体外放射治疗(EBRT)的经验，我们知道根据患者的具体情况开出处方 吸收剂量会带来更好的患者结果。实现这种个性化所需的关键技术是快速， 准确的全身特定于患者的剂量测定。在之前的第二阶段SBIR中，Voximeter能够将我们的 完全基准和IP保护的蒙特卡罗剂量学算法成为经济有效的RPT治疗 计划解决方案(TorchTM)，通过添加图像配准、轮廓传播和 基于体素的药代动力学(PK)建模。Torch的设计目的是模拟任何 药物类别(如小分子、多肽、抗体)和任何放射性核素，有效地增加了一支“瑞士军队” 将刀具放入临床医生的RPT工具包中。利用RPT代理发出的伽马射线，Torch可以 计算在全身器官和肿瘤的每个体素内沉积了多少辐射能量。 体素水平全身患者特定剂量测量的估计与反应的相关性比 目前临床上可用的标准摄取值(SUV)信息。使用这种方法，Voximeter已经 开发了快速、准确、RPT的治疗计划解决方案，旨在以极其准确的方式告知临床医生 基于体素的蒙特卡罗(MC)剂量测量只需几分钟即可完成。此外，Vox还开发了一款图形处理器 加速蒙特卡罗SPECT重建算法，导致剂量估计不同于 传统的重建算法至少提高了25%，这被视为具有临床意义。此信息 将支持个性化安全治疗剂量的临床决策。此RPT剂量测定工作流程的自动化 对于希望在临床上实施剂量学指导治疗的医疗系统来说尤其重要 实践，但目前在专业知识和资源方面装备不足，无法执行高级剂量测量 RPT。通过实现自动化，Vox最终将通过提供个性化的 针对转移性癌症的治疗，在许多情况下比 化疗。该提案将实现的具体目标是：(1)开发自动化 使用人工智能对器官、风险和肿瘤进行分割的工具，(2)开发扫描仪校准， 部分体积校正和蒙特卡罗重建，以确保准确的SPECT成像，以及(3)自动化， 实施并验证端到端的临床火炬工作流。这些目标的成功实现将有助于 打破商业化障碍，加速Torch的市场应用。

PROJECT SUMMARY/ABSTRACT The Radiopharmaceutical Therapy (RPT) market is projected to surpass the Technitium-99m market by 2025, with an emphasis on metastatic prostate cancer, neuroendocrine tumors, and lymphoma. Driven by well- tolerated treatments and fewer side-effects, experts have estimated 150 new theranostic centers will be needed in the U.S. to deliver an estimated 50,000–200,000 treatment cycles/year. Currently, all RPT treatments administer a standard therapeutic dose despite unique patient physiology and pharmacokinetics. From experience with external beam radiation therapy (EBRT), we know that patient-specific prescriptions based on absorbed dose leads to better patient outcomes. A key technology required to enable this personalization is fast, accurate, whole-body patient-specific dosimetry. In a prior Phase II SBIR, Voximetry was able to integrate our fully benchmarked and IP-protected Monte Carlo dosimetry algorithm into a cost-effective RPT treatment planning solution (TorchTM) by adding additional features such as image registration, contour propagation, and voxel-based pharmacokinetic (PK) modeling. Torch was designed to model uptake and clearance routes of any drug class (e.g., small molecules, peptides, antibodies) and any radionuclide, effectively adding a ‘swiss army knife’ tool into the clinician’s RPT toolkit. Leveraging the gamma rays emitted by the RPT agent, Torch can calculate how much radiation energy is deposited within each voxel of organs and tumors throughout the body. Estimates of voxel-level whole-body patient-specific dosimetry are better correlated with response than the standard uptake value (SUV) information that is clinically available today. Using this approach, Voximetry has developed a fast, accurate, RPT treatment planning solution aimed to inform clinicians with extremely accurate voxel-based Monte Carlo (MC) dosimetry in a matter of minutes. Additionally, Vox has developed a GPU accelerated Monte Carlo SPECT reconstruction algorithm that leads to dosimetry estimates which differ from conventional reconstruction algorithms by at least 25%, which is seen as clinically significant. This information will support clinical decisions to personalize safe therapeutic doses. Automation of this RPT dosimetry workflow is especially important for healthcare systems that would like to implement dosimetry-guided therapy in clinical practice but are currently ill-equipped in terms of expertise and resources to perform advanced dosimetry for RPT. By enabling automation, Vox will ultimately benefit cancer patients by making available a personalized treatment that targets metastatic cancer that in many cases is more efficacious and has fewer side effects than chemotherapy. The specific aims that will be accomplished in the proposal are to (1) Develop automated segmentation tools for organs and risk and tumors using artificial intelligence, (2) develop scanner calibration, partial volume correction, and Monte Carlo reconstruction to ensure accurate SPECT imaging, and (3) automate, implement, and validate end-to-end clinical Torch workflow. The successful completion of these aims will help break down barriers to commercialization and accelerate market adoption of Torch.