Advanced Algorithmic Acceleration and System Modeling for Low-Dose CT Imaging

低剂量 CT 成像的先进算法加速和系统建模

• 批准号: 8549377
• 负责人: Jeffrey Brokish
• 金额: $ 55.56万
• 依托单位: INSTARECON, INC.
• 依托单位国家: 美国
• 财政年份: 2012
• 资助国家: 美国
• 起止时间: 2012-08-01 至 2015-07-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/8549377
• 关键词: Acceleration; Algorithms; Artificial cardiac pacemaker; Biological Models; Childhood; Clinical; Data; Detection; Development; Diagnostic; Dose; Dose-Limiting; Family; Goals; Grant; Health Care Costs; Heart Diseases; Hospitals; Housing; Image; Image-Guided Surgery; Implant; Industry; Left; Malignant neoplasm of lung; Marketing; Measures; Metals; Metric System; Modality; Modeling; Modification; Morphologic artifacts; Noise; Patients; Performance; Phase; Physics; Process; Prosthesis; Public Health; Radiation; Reading; Relative (related person); Resolution; Risk; Running; Scanning; Small Business Innovation Research Grant; Source; System; Testing; Time; Translating; United States National Institutes of Health; Update; Work; X-Ray Computed Tomography; advanced system; base; cancer radiation therapy; cost; data acquisition; detector; improved; lung cancer screening; new technology; patient population; prevent; prototype; radiologist; reconstruction; screening; statistics

DESCRIPTION (provided by applicant): With the increased use of x-ray CT, the development of a market for CT screening exams, and the imaging of younger patients, there is a growing concern about the public health risk caused by the radiation dose delivered by x-ray CT. The reduction of this dose has therefore taken on increased importance, as evidenced by the recent NIH Summit on Managing Dose in CT with the mandate of achieving the routine sub-millisievert CT exam. Iterative reconstruction algorithms are a key part in accomplishing this goal, producing high-quality images from low-dose data by incorporating detailed models of the physics and statistics of the data acquisition process. Iterative algorithms based on these system models are beginning to enter the marketplace, but currently these algorithms suffer from three main limitations: (i) they are a very expensive add-on; (ii) they leave out detailed modeling of the physics, thus limiting the available dose reduction; and (iii) they are 10 - 100 times slower than standard reconstruction, preventing their use as a default for routine scans. The key to fully enabling iterative algorithms is acceleration of the backprojection and reprojection computational bottleneck, which is accomplished through the use of InstaRecon's fast hierarchical backprojection/reprojection operators. Accelerating the iterative algorithm enables it to run on a less expensive platform, delivering fast reconstruction rates, and opens the door to incorporation of other system modeling, allowing for further image quality improvement and dose reduction. Thus, low-dose imaging and iterative reconstruction can move from a high-end option to the default scanning mode for a wide range of CT scanner hardware. The overall goal of this SBIR project is to accelerate iterative reconstruction rates even further and incorporate additional system models to improve dose and artifact reduction capabilities. The system acceleration will be achieved through algorithmic modifications to the hierarchical operators and the iterative reconstruction loop itself. Additional system modeling wil be introduced at a reduced computational cost through incorporation into the hierarchical operators themselves, providing advanced, accelerated system models. The resulting system will be faster than existing iterative reconstruction platforms, run on less expensive hardware, with additional reduction in dose and artifact levels. Benefits of the new technology will include superior low-dose performance in dose-critical applications such as pediatric, screening for lung cancer or heart disease, and interventional imaging, and significant improvement in diagnostic quality of CT scans of large patients, or of patients with prosthetic implants or cardiac pacemakers. Moreover, this project will help make iterative algorithm-based low-dose imaging a common scanning modality, reducing the burden of CT x-ray exposure for the patient population at large.

描述(申请人提供)：随着X光CT使用的增加，CT筛查检查市场的发展，以及年轻患者的成像，人们越来越关注由X光CT提供的辐射剂量所造成的公共健康风险。因此，减少这一剂量变得越来越重要，最近的NIH CT剂量管理峰会证明了这一点，该峰会的任务是实现常规的亚毫西弗CT检查。迭代重建算法是实现这一目标的关键部分，通过结合详细的物理模型和数据采集过程的统计数据，从低剂量数据生成高质量的图像。基于这些系统模型的迭代算法开始进入市场，但目前这些算法受到三个主要限制：(I)它们是非常昂贵的附加组件；(Ii)它们遗漏了物理的详细建模，从而限制了可用的剂量减少；以及(Iii)它们比标准重建慢10-100倍，使其无法作为常规扫描的默认使用。完全启用迭代算法的关键是加速反投影和重投影计算瓶颈，这是通过使用InstaRecon的快速分层反投影/重投影操作符来实现的。加速迭代算法使其能够在更便宜的平台上运行，提供快速重建速率，并为整合其他系统建模打开了大门，允许进一步提高图像质量和减少剂量。因此，低剂量成像和迭代重建可以从高端选项转变为广泛的CT扫描仪硬件的默认扫描模式。这个SBIR项目的总体目标是进一步加快迭代重建速度，并纳入更多的系统模型，以提高剂量和伪影减少能力。系统加速将通过对分层算子和迭代重建循环本身的算法修改来实现。额外的系统建模将通过并入分层运算符本身，以降低计算成本的方式引入，从而提供先进的、加速的系统模型。由此产生的系统将比现有的迭代重建平台更快，运行在更便宜的硬件上，并进一步减少剂量和伪影水平。这项新技术的好处将包括在儿科、肺癌或心脏病筛查和介入性成像等剂量关键型应用中的卓越低剂量性能，以及对大型患者、或接受假体植入或心脏起搏器的患者的CT扫描诊断质量的显著提高。此外，该项目将有助于使基于迭代算法的低剂量成像成为一种常见的扫描方式，从而减轻广大患者群体接受CT X射线照射的负担。