A GPU-cloud based Monte Carlo simulation platform for National Particle Therapy Research Center

国家粒子治疗研究中心基于GPU云的蒙特卡罗模拟平台

• 批准号: 8811782
• 负责人: Steve Bin Jiang
• 金额: $ 21.37万
• 依托单位: UT SOUTHWESTERN MEDICAL CENTER
• 依托单位国家: 美国
• 财政年份: 2015
• 资助国家: 美国
• 起止时间: 2015-02-10 至 2017-01-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/8811782
• 关键词: Address; Anatomy; Area; Carbon; Clinical; Cloud Computing; Code; Communities; Conflict (Psychology); Databases; Development; Dose; Ensure; Geometry; Goals; Hour; Imagery; Internet; Language; Measurement; Microscopic; Mission; Modeling; Monte Carlo Method; Nature; Physics; Pilot Projects; Play; Process; Radiation therapy; Radiobiology; Research; Research Activity; Research Personnel; Resources; Role; Running; Sampling; Services; Staging; Structure; System; Systems Development; Techniques; Technology; Test Result; Testing; Therapeutic Studies; Time; Uncertainty; Validation; base; cloud based; flexibility; handheld mobile device; laptop; novel; particle; particle beam; particle physics; particle therapy; physical process; prototype; simulation; statistics; success; therapy design; tool; treatment planning; usability; user-friendly; virtual; web interface

Project Summary Monte Carlo (MC) simulation is a valuable tool for radiation therapy. Particularly for particle beam radiation therapy (PBRT), its remarkable value has been well recognized. Examples include, but not limited to, accurately calculating dose distributions that are highly sensitive to treatment geometry and anatomy, reducing range uncertainty, developing novel treatment verification techniques, capturing radiobiological effects from the microscopic level, and designing treatment facility. Hence, researchers are eager to have a fast, robust, and easy-to-use MC system in their studies. Yet, there are two main difficulties to use current available MC packages for PBRT, namely low computational efficiency and highly required user expertise. The conflicts between the great desire of using MC and the difficulties of using it have impeded research and clinical activities in PBRT to significantly. As part of the planning process for National Particle Therapy Research Center (NPTRC), we propose in this pilot project a highly accurate, efficient, yet user-friendly centralized MC simulation system using novel graphics-processing unit (GPU) and cloud-computing technologies. Different from conventional MC packages running on the user's end, our system remotely resides in a cloud inside NPTRC and provides MC simulation services to PBRT researchers though standard web browsers. While our long-term goal is to deliver novel MC simulations to facilitate the establishments of NPTRC and its future research activities, as well as to service the entire PBRT community, the goal of this pilot project is to initiate efforts toward the long-term goal by developing and validating a prototype system focusing on particle beam dose calculations to demonstrate feasibility and impacts. The deliverability of this project has been clearly demonstrated by mature technologies and our extensive preliminary studies. The strong research team, particularly the integration of Dr. Parodi for particle physics modeling, also ensures success. Our goal will be accomplished by pursuing two specific aims (SAs): (1) System developments: develop web interface, physics database, and core GPU-based MC simulation codes. (2) System validations: Comprehensively validate the computational accuracy of our system and test its efficiency. Perform end-to-end functionality test in a representative research scenario. This pilot project fits into the overall plan for the proposed NPTRC facility. (1) Being an integral component of NPTRC, it will play a critical role for the planning stage by offering virtual yet realistic simulations of different clinical, physical, and technical scenarios. In the long run, our system will greatly expand NPTRC's research capacity and hence significantly contribute to the establishments of its leading role in PBRT field. (2) Our system service PBRT field with high quality MC simulations. Continuous developments will add much more features to address needs from different research aspects. This is aligned with the NPTRC's mission of providing resources for researchers to investigate important problems in PBRT.

项目摘要 蒙特卡罗（MC）模拟是一种有价值的工具，放射治疗。特别是对于粒子束辐射 PBRT的治疗，其显着的价值已得到充分的认可。实例包括但不限于， 准确计算对治疗几何形状和解剖结构高度敏感的剂量分布， 范围的不确定性，开发新的治疗验证技术，捕获放射生物学效应， 微观层面，设计处理设施。因此，研究人员渴望有一个快速，强大， 在他们的研究中使用简单易用的MC系统。然而，使用当前可用的MC存在两个主要困难 PBRT的软件包，即计算效率低，对用户专业知识的要求很高。的冲突 在使用MC的巨大愿望和使用困难之间，阻碍了研究和临床 在PBRT的活动，以显着。作为国家粒子治疗研究规划过程的一部分 中心（NPTRC），我们在这个试点项目中提出了一个高度准确，高效，但用户友好的集中MC 这是一个使用新型图形处理单元（GPU）和云计算技术的模拟系统。不同 从传统的MC包运行在用户端，我们的系统远程驻留在云内 NPTRC和提供MC模拟服务PBRT研究人员通过标准的Web浏览器。虽然我们的 长期目标是提供新的MC模拟，以促进NPTRC的建立及其未来 研究活动，以及服务整个PBRT社区，这个试点项目的目标是启动 通过开发和验证聚焦于粒子束的原型系统，努力实现长期目标 剂量计算，以证明可行性和影响。该项目的可交付性已被明确 成熟的技术和我们广泛的初步研究证明。强大的研究团队， 特别是Parodi博士在粒子物理建模方面的整合，也确保了成功。我们的目标将是 通过追求两个具体目标（SA）来实现：（1）系统开发：开发Web界面，物理 数据库，以及核心的基于GPU的MC仿真代码。(2)系统验证：全面验证 我们的系统的计算精度和测试其效率。执行端到端功能测试 典型的研究场景。该试点项目符合拟议的NPTRC设施的总体计划。（一） 作为NPTRC的一个组成部分，它将通过提供虚拟然而， 不同临床、物理和技术场景的逼真模拟。从长远来看，我们的系统将 大大扩大NPTRC的研究能力，从而大大有助于其建立 在PBRT领域发挥主导作用。(2)我们的系统为PBRT领域提供高质量的MC模拟。连续 发展将增加更多的功能，以满足不同研究方面的需要。这是对齐的 NPTRC的使命是为研究人员提供研究PBRT中重要问题的资源。