Deep Radiomic Analysis of Coronary Heart Disease with Lung Screening CT

通过肺部筛查 CT 进行冠心病的深度放射组学分析

基本信息

批准号：
10009810
负责人：
Mannudeep Karanvir Singh Kalra
金额：
$ 36.97万
依托单位：
RENSSELAER POLYTECHNIC INSTITUTE
依托单位国家：
美国
项目类别：
财政年份：
2019
资助国家：
美国
起止时间：
2019-09-15 至 2021-08-31
项目状态：
已结题

来源：
https://reporter.nih.gov/project-details/10009810
关键词：
Accounting Address Affect Algorithms Angiography Body fat Calcium Cardiovascular Diseases Cardiovascular system Cessation of life Communities Comorbidity Consensus Coronary Coronary heart disease Data Dietary Practices Evaluation Funding Goals Graph Health Healthcare Image Life Style Lung Lung CAT Scan Machine Learning Malignant Neoplasms Malignant neoplasm of lung Methods Morbidity - disease rate National Heart, Lung, and Blood Institute Obesity Patient Education Patients Performance Population Radiation Radiation Dose Unit Recording of previous events Recurrence Reference Standards Reporting Risk Risk Factors Scanning Semantics Series Smoker Smoking Source Stress Techniques Time Tobacco use United States United States Centers for Medicare and Medicaid Services Validation Vendor X-Ray Computed Tomography calcification cardiovascular disorder risk cardiovascular risk factor clinical practice cohort coronary artery calcification cost deep learning deep neural network follow-up healthy lifestyle high risk high risk population image reconstruction improved innovation low-dose spiral CT lung cancer screening mortality mortality risk muscle form prevent prognostic prognostic performance radiologist radiomics reconstruction screening spatiotemporal task analysis unhealthy lifestyle

项目摘要

Cardiovascular diseases (CVDs) caused 17.7 million deaths in 2015, accounting for one third of all deaths globally. Amongst 8.8 million cancer-deaths in 2015, lung cancer contributed to 1.69 million deaths worldwide. In the United States, lung cancer will claim 154,050 lives in 2018, representing a quarter of all cancer deaths. Both CVDs and lung cancer share several risk factors, including unhealthy dietary patterns, obesity, and tobacco use. Low-dose Computed Tomography (LDCT), an effective lung cancer screening technique, reduces the cancer-related mortality rate by 20% in those with history of heavy smoking. Such subjects have substantially higher risk of CVDs than lung cancer. Indeed, most subjects undergoing lung cancer screening with LDCT have an intermediate to high risk for CVDs. Concomitant CVD screening will have profound benefits for these subjects at no additional time, cost, or radiation dose to the patient. The goal of the project is to develop an automated workflow of cardiovascular morbidity and mortality risk (CARMOR) evaluation with lung cancer screening LDCT data to provide high-risk subjects a “radiation-free” CVD screening without cost or time constraints. Our central hypothesis is that deep radiomics from LDCT follow-up scans and across different reconstructions can effectively improve the performance of CARMOR estimation. To validate our hypothesis, we will perform radiomics analysis on CARMOR with lung cancer screening LDCT scans, and then further enhance the CARMOR features with complementary image reconstructions from the same scan and serial LDCT scans for optimized prognostic performance. The CARMOR quantification methods in lung screening LDCT will be validated retrospectively against multiple subspecialty radiologists with Coronary Computed Tomography Angiogram (CCTA) as the standard of reference. The project requires us to perform analyses over serial LDCT exams and different reconstruction techniques and settings. Therefore, major innovations and improvements need to be made for this challenging deep radiomic analysis task. Since the risk factors of CVDs and lung cancer are closely related with lifestyle, the findings of the project can also be used to create teachable moments for patients to change towards healthier lifestyles. In context of the Center for Medicare & Medicaid Services (CMS) reimbursement of LDCT for lung cancer screening and stress on the importance of considering co-morbidities, our timely project will have a significant impact on the healthcare of a large population. At the end of the project, we will have developed an automated workflow of assessing CARMOR for lung screening LDCT with zero radiation added. We will also have acquired study results to enable personalized patient education to address the long-term health crisis caused by unhealthy lifestyles. Our studies will lead to new understanding of deep radiomics in CVD screening and the impact of LDCT reconstructions in clinical practice.

2015年，心血管疾病（CVD）导致1770万人死亡，占总死亡人数的三分之一 2015 年，全球 880 万人因癌症死亡，其中 169 万人死于肺癌。在美国，2018 年肺癌将夺走 154,050 人的生命，占所有癌症死亡人数的四分之一。心血管疾病和肺癌都有几个共同的危险因素，包括不健康的饮食模式、肥胖和烟草低剂量计算机断层扫描 (LDCT) 是一种有效的肺癌筛查技术，可减少肺癌筛查。在有大量吸烟史的受试者中，癌症相关死亡率显着降低20%。事实上，大多数接受 LDCT 肺癌筛查的受试者患有 CVD 的风险高于肺癌。 CVD 中度至高风险的伴随 CVD 筛查将为这些受试者带来深远的益处。无需额外的时间、成本或对患者的辐射剂量。该项目的目标是开发心血管发病和死亡风险的自动化工作流程 (CARMOR) 利用肺癌筛查 LDCT 数据进行评估，为高危受试者提供“无辐射”CVD 我们的中心假设是 LDCT 随访的深入放射组学。扫描和跨不同重建可以有效提高CARMOR估计的性能。为了验证我们的假设，我们将通过肺癌筛查 LDCT 对 CARMOR 进行放射组学分析扫描，然后通过互补图像重建进一步增强 CARMOR 特征相同的扫描和串行 LDCT 扫描可优化预后性能 CARMOR 量化方法。在肺部筛查中，LDCT 将针对多名冠状动脉亚专科放射科医生进行回顾性验证计算机断层扫描血管造影（CCTA）作为参考标准该项目要求我们执行。对连续 LDCT 检查和不同重建技术和设置的分析因此是主要的。对于这项具有挑战性的深度放射组学分析任务，需要进行创新和改进。由于心血管疾病和肺癌的危险因素与生活方式密切相关，该项目的研究结果可以也可用于创造教育时刻，让患者转向更健康的生活方式。医疗保险和医疗补助服务中心 (CMS) 报销用于肺癌筛查和压力的 LDCT 关于考虑合并症的重要性，我们及时的项目将对医疗保健产生重大影响在项目结束时，我们将开发一个自动化的评估工作流程。 CARMOR 用于肺部筛查 LDCT，添加零辐射我们还将获得研究结果以实现这一目标。个性化的患者教育，以解决由不健康的生活方式引起的长期健康危机。将带来对 CVD 筛查中的深度放射组学以及 LDCT 重建的影响的新认识临床实践。