AUTOMATIC QUANTITATIVE CT IMAGING OF PERICARDIAL FAT: A NOVEL ISCHEMIA PREDICTOR

描述(申请人提供)：在美国，每年有100万人经历心脏病发作或心脏性猝死。这些患者中有很大一部分以前没有任何症状，但患有无症状的心脏病(缺血)，这可能会在任何时候导致心脏病发作。目前还没有可靠的筛查方法来识别可能患有静止性心脏病并因此有“心脏病发作”风险的人。 我们寻求从心脏的非侵入性计算机断层扫描(CT)(通常用于大多数成像中心)中提取额外的数据，这将更好地识别出易受突发心脏病发作影响的患者。这些图像是在没有造影剂的情况下获得的，作为测量冠状动脉钙的常见筛查测试，冠状动脉钙是冠状动脉疾病的另一个临床标志。我们将证明，通过在这些心脏CT图像中量化心脏周围的心包脂肪，可以获得关于心血管风险的额外信息。虽然这一信息存在于今天获得的心脏CT扫描中，但在临床实践中它被忽视了，因为目前还没有工具来自动和可靠地测量它，也没有关于它对特定患者的重要性的信息。 我们将开发全自动、准确的新计算机软件，用于从常规评估冠状动脉钙离子的心脏CT图像中定量心包脂肪。我们计划通过将稳健和有效的算法与解剖学知识相结合，在机器学习中实现完全自动化和准确性。我们的目标是证明，开发的自动软件在量化心包脂肪方面将与经验丰富的成像医生一样准确，更具重复性，并将在很短的时间内完成测量。这样的软件可以立即在全球范围内提供给标准成像工作站。这将首次允许将其用作心血管风险评估的标准实用成像工具，而不需要额外的时间、成像或对患者的辐射风险。 我们还将展示，与其他临床心血管风险因素和来自血液样本的信息(如胆固醇、葡萄糖和C反应蛋白)相比，这种新的定量信息具有重要的附加价值，以及在CT扫描中看到的患者冠状动脉中的钙量，并确立其在预测无症状心脏病方面的意义。 与公共健康相关：这项新研究将使人们能够准确预测谁更有可能患有心脏病，而心脏病最终可能导致心脏病发作。它将允许医生识别可以为哪些患者开适当的治疗以避免心脏病发作，或者可以建议进行更广泛的测试，以确认或排除心脏病。

DESCRIPTION (provided by applicant): Every year, 1 million people in the US will experience a heart attack or sudden cardiac death. A large percentage of these patients have no prior symptoms of any kind but suffer from silent heart disease (ischemia), which may cause a heart attack at any time. Currently there is no reliable screening method to identify people who may have silent heart disease and therefore are at risk of "heart attack". We seek to extract additional data from a non-invasive Computed Tomography (CT) scan of the heart (commonly used in most imaging centers), which will better identify patients who are "vulnerable" to sudden heart attack. These images are acquired without contrast agents, as a common screening test for measuring coronary calcium, another clinical marker of coronary artery disease. We will prove that additional information about cardiovascular risk can be obtained by quantifying pericardial fat surrounding the heart in these cardiac CT images. Although this information exists in the cardiac CT scans acquired today, it is ignored in clinical practice because currently there are no tools to measure it automatically and reliably, and no information regarding its significance for a given patient. We will develop fully automated, accurate, new computer software for quantitation of pericardial fat from cardiac CT images acquired for routine assessment of coronary calcium. We plan to achieve complete automation and accuracy, by applying a combination of robust and effective algorithms in machine learning with anatomical knowledge. We aim to prove that the developed automatic software will be as accurate, and more reproducible than experienced imaging physicians in quantifying pericardial fat, and will accomplish the measurement in a fraction of the time. Such software could be immediately made available worldwide to standard imaging workstations. For the first time, this will allow its use as a standard practical imaging tool for cardiovascular risk assessment, without requiring additional time, imaging or radiation risk to the patient. We will also show that this new quantitative information has important additional value compared to other clinical cardiovascular risk factors and information from blood samples (such as cholesterol, glucose and CRP), and the amount of calcium in the patient's coronary arteries seen on CT scans, and establish its significance in the prediction of silent heart disease. PUBLIC HEALTH RELEVANCE: This new research will allow accurate prediction of who might be more likely to have heart disease which may ultimately cause a heart attack. It will allow doctors to identify patients for whom appropriate treatment could be prescribed to avoid a heart attack, or more extensive testing, could be recommended, to confirm or rule out heart disease.