Novel Algorithms for Reducing Radiation Dose of CT Perfusion

项目摘要/摘要 X射线计算机断层扫描(CT)已越来越多地应用于医疗诊断，目前已达到更多 在美国，每年有超过1亿次CT扫描。CT的使用越来越多，引发了人们对 辐射剂量对患者的影响。据估计，每2000次CT扫描将导致一种未来的癌症，即， 每年从1亿例CT扫描中发现5万例未来癌症。CT脑灌注成像(CTP)是一种应用广泛的 应用影像技术评价卒中和脑血管疾病的血流动力学变化。 然而，CTP对患者来说涉及到高辐射剂量，因为CTP扫描要重复40次左右 在相同的解剖位置，以捕捉造影剂的完整通道。几种技术 已应用于减少CTP扫描中的辐射剂量，包括减少管电流和管 以及使用迭代重建(IR)等降噪技术。然而， 现有CTP扫描的合成辐射剂量仍然显著高于标准头部CT扫描。 红外光谱技术在CTP中的应用受到了很大的限制，其复杂性和计算量都很大 用于处理影响临床工作流程的多个CTP图像。在第一阶段的STTR项目中，我们 介绍了一种基于k空间加权图像对比度(KWIC)的低剂量CTP成像方法 重建算法。我们在CTP模体和临床CTP数据集中进行了全面的评估， 并证明了KWIC算法能够通过以下方式减少现有CTP技术的辐射剂量 75%，而不影响图像质量和量化的准确性(即，第一阶段STTR的里程碑)。 然而，最初的KWIC算法需要在预先指定的旋转角下快速切换脉冲X射线-a 硬件功能尚未由商业CT供应商实施。为了解决这一限制，我们 最近引入了KWIC算法的一个变体，称为k空间加权图像平均(KWIA)，该算法 保持较高的空间和时间分辨率以及低剂量(~75%剂量)CTP数据的图像质量 减少)以与标准CTP扫描相媲美。最重要的是，KWIA不需要 对现有CT硬件的修改，计算简单和快速，因此具有低门槛 市场渗透率。第二阶段STTR项目的目的是进一步优化和验证KWIA 降低CTP扫描辐射剂量~75%的算法，同时保持图像质量和 CTP模体、临床CTP数据和动物研究中的量化准确性。我们将进一步发展 创新的基于深度学习(DL)的算法，以解决KWIA中的潜在运动和其他伪影，以及 通过与CT供应商合作，将开发的算法商业化。

Project Summary/Abstract X-ray computed tomography (CT) has been increasingly used in medical diagnosis, currently reaching more than 100 million CT scans every year in the US. The increasing use of CT has sparked concern over the effects of radiation dose on patients. It is estimated that every 2000 CT scans will cause one future cancer, i.e., 50,000 cases of future cancers from 100 million CT scans every year. CT brain perfusion (CTP) is a widely used imaging technique for the evaluation of hemodynamic changes in stroke and cerebrovascular disorders. However, CTP involves high radiation dose for patients as the CTP scan is repeated on the order of 40 times at the same anatomical location, in order to capture the full passage of the contrast bolus. Several techniques have been applied for radiation dose reduction in CTP scans, including reduction of tube current and tube voltage, as well as the use of noise reduction techniques such as iterative reconstruction (IR). However, the resultant radiation dose of existing CTP scans is still significantly higher than that of a standard head CT scan. The application of IR techniques in CTP is very limited due to the high complexity and computational burden for processing multiple CTP images that impairs clinical workflow. During the Phase 1 STTR project, we introduced a novel low dose CTP imaging method based on the k-space weighted image contrast (KWIC) reconstruction algorithm. We performed thorough evaluation in both a CTP phantom and clinical CTP datasets, and demonstrated that the KWIC algorithm is able to reduce the radiation dose of existing CTP techniques by 75% without affecting the image quality and accuracy of quantification (i.e., Milestone of Phase 1 STTR). However, the original KWIC algorithm requires rapid-switching pulsed X-ray at pre-specified rotation angles – a hardware capability yet to be implemented by commercial CT vendors. In order to address this limitation, we recently introduced a variant of the KWIC algorithm termed k-space weighted image average (KWIA) that preserves high spatial and temporal resolutions as well as image quality of low dose CTP data (~75% dose reduction) to be comparable to those of standard CTP scans. Most importantly, KWIA does not require modification of existing CT hardware and is computationally simple and fast, therefore has a low barrier for market penetration. The purpose of the Phase 2 STTR project is to further optimize and validate the KWIA algorithm for reducing radiation dose of CTP scans by ~75% while preserving the image quality and quantification accuracy in CTP phantom, clinical CTP data and animal studies. We will further develop innovative deep-learning (DL) based algorithms to address potential motion and other artifacts in KWIA, and commercialize the developed algorithms by collaborating with CT vendors.