High-performance computing with graphics hardware for medical physics applications

用于医学物理应用的具有图形硬件的高性能计算

• 批准号: 355493-2013
• 负责人: Després, Philippe
• 金额: $ 1.97万
• 依托单位: Université Laval
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2014
• 资助国家: 加拿大
• 起止时间: 2014-01-01 至 2015-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=549743
• 关键词: performance; computing; graphics; hardware; medical

A vast array of numerical methods are used to solve problems in Medical Physics, ranging from purely analytical approaches to Monte Carlo techniques. These numerical methods are essential for instance in dose calculation in radiation oncology and in all image processing tasks, including tomographic image reconstruction in radiology and nuclear medicine. The accuracy of numerical solutions typically grows with the complexity of the model used to describe the physics of the problem. Monte Carlo techniques for example provide the most accurate solutions to dose calculation problems because they are based on fundamental physics processes. However, the computational burden and long execution times associated with physics-rich models are typically not compatible with clinical applications, which often require near real-time execution. In order to deploy physics-rich, more accurate algorithms in clinical environments, my research group has developed high-performance computing approaches based on Graphics Processing Units (GPUs). We have developed a recognized expertise in adapting Medical Physics problems to these massively parallel devices, using innovative strategies to significantly accelerate advanced dose calculation and tomographic reconstruction algorithms. Building upon these previous achievements made mostly in Radiation Oncology, we plan to continue this research program to further develop innovative applications in Radiology and Nuclear Medicine. The overall objective is to integrate advanced physics models into Medical Physics applications in order to reach better, more accurate solutions in acceptable timeframes.

大量的数值方法被用于解决医学物理学中的问题，范围从纯粹的分析方法到蒙特卡罗技术。这些数值方法是必不可少的，例如在放射肿瘤学的剂量计算和所有图像处理任务，包括放射学和核医学的断层成像图像重建。数值解的准确性通常随着用来描述问题物理性质的模型的复杂性而增长。例如，蒙特卡罗技术为剂量计算问题提供了最精确的解决方案，因为它们基于基本的物理过程。然而，与物理丰富的模型相关的计算负担和长时间的执行时间通常与临床应用不兼容，临床应用通常需要接近实时的执行。为了在临床环境中部署丰富的物理，更准确的算法，我的研究小组开发了基于图形处理单元（gpu）的高性能计算方法。我们已经开发出公认的医学物理问题适应这些大规模并行设备的专业知识，使用创新的策略来显著加快高级剂量计算和层析成像重建算法。基于这些先前在放射肿瘤学方面取得的成就，我们计划继续这项研究计划，进一步开发放射学和核医学方面的创新应用。总体目标是将先进的物理模型集成到医学物理应用程序中，以便在可接受的时间范围内获得更好、更准确的解决方案。