GPU-Based Monte Carlo Software for Computing CT Imaging Doses

描述（由申请人提供）：该项目的目标是展示一种新的蒙特卡罗辐射模拟软件包，用于低成本的台式高性能计算机，这将大大减少计算机断层扫描（CT）成像剂量计算的周转时间。x射线是最古老和最广泛使用的诊断成像方式。然而，关于辐射致癌效应的潜在风险的争议一直围绕着三维体积CT身体成像的临床应用。然而，经验表明，现有的患者CT剂量计算工具不足以进行患者特异性风险评估、扫描仪优化、方案比较和事故调查。我们提出了一种创新的方法，利用通用图形处理单元（gpgpu），以前所未有的低价格从大规模并行处理器中获取巨大的能量。我们组建了这个由核工程、蒙特卡罗辐射输运理论、医学物理、计算机科学、临床放射学和CT扫描仪设计等多学科专家组成的团队，以实现以下具体目标：开发一个新的蒙特卡罗软件包，专门为新兴的混合CPU/GPU并行计算平台设计和优化，并验证该软件。2. 将该软件与GE光速CT扫描仪模型和可变形患者幻影库集成。3. 展示和评估新的蒙特卡罗计算工具在典型诊断CT扫描方案中的临床效益。4. 建立全国临床放射界蒙特卡洛并行计算资源中心

DESCRIPTION (provided by applicant): This project's goal is to demonstrate a new Monte Carlo radiation simulation software package, for a low-cost desktop high-performance computer, that will significantly reduce the turn-around time for computed tomography (CT) imaging dose calculations. X-ray is the oldest and most widely used diagnostic imaging modality. However, controversy about the potential risk of radiation-induced carcinogenic effects has always surrounded the clinical use of 3D volumetric CT body imaging. However, experiences suggest that existing patient CT dose computational tools are insufficient for patient- specific risk assessment, scanner optimization, protocol comparison, and accident investigation. We propose an innovative approach which takes advantage of general-purpose graphics processing units (GPGPUs) that harvest the enormous power from massively parallel processors at unprecedented low prices. We have assembled this multidisciplinary team of experts from nuclear engineering, Monte Carlo radiation transport theory, medical physics, computer science, clinical radiology, and CT scanner design to achieve the following Specific Aims: 1. To develop a new Monte Carlo software package that is specifically designed and optimized for the emerging hybrid CPU/GPU parallel computing platforms and to validate the software. 2. To integrate the software with GE LightSpeed CT scanner models and a library of deformable patient phantoms. 3. To demonstrate and evaluate clinical benefits of the new Monte Carlo computing tool for typical diagnostic CT scanning protocols. 4. To establish a national resource center of Monte Carlo parallel-computing for the clinical radiological community