Multi-Center Implementation and Validation of Efficient Magnetic Resonance Imaging and Analysis of Atherosclerotic Disease of the Cervical Carotid

摘要： 在过去的几十年中，许多研究已经在用于磁共振成像的MR方法中产生了实质性的创新。 颅外颈动脉粥样硬化的特征。在理想条件下通过这些创新获得的图像 这些条件为临床医生提供了关于动脉壁疾病的丰富信息， 对这种潜在疾病进行适当管理是非常必要的。尽管如此， 这项技术还没有成为常规的临床医疗设备。事实上，由于需要 钆基造影剂（GBCA），长的检查时间（通常约45分钟，以获得 在5或6个必要的序列中的多个对比），以及解释多个 对比MRI大多数从业者在评估风险时仍然恢复到直径狭窄的简化度量。后 经过多年的集体调查，合作研究这一提案的调查人员联盟认为， 现在是时候解决这些剩余的局限性，并最终改变临床模式。 过度假设：为了实现颈动脉疾病管理的巨大潜力， 需要高效且易于使用MRI技术。我们的假设是， 使用多参数非对比MRI序列结合最新的高信噪比（SNR） 颈部形状特定（NSS）射频线圈和创新的机器学习（深度神经网络）分析方法。 目标1：我们将在我们的5个参与中心安装相同的RF线圈、MRI序列和方案， 并严格测试所有中心的测量精度和图像质量的再现性。目的 2：我们将开发、训练和验证一个用户友好的深度学习神经网络系统，用于定量分析。 分析被认为存在于易损动脉粥样硬化斑块中的几种关键成分。目标3： 我们将对一组颈动脉疾病受试者进行分析，以确定 定量测量以及与组织病理学相比表征的准确性。虽然 我们将在一个高技能的学术网络中开发和测试图像质量，再现性和可靠性。 中心，我们将设计这些方法适用于社区医院的设置。结束时 在这个项目中，我们建议有一个可用于后续调查的综合解决方案，例如： 药物干预在改变斑块组成中的作用;研究 随着时间的推移，未经治疗的动脉粥样硬化疾病的特征;最终，调查 预测有害的结果，并可用于改善这一人群的干预策略。 成功完成后，将提供RF线圈和MRI序列以及分析方法， 其他成像中心的方式，最终改变了诊断和管理的范式， 治疗颈动脉粥样硬化疾病。

Abstract: Numerous investigations over the past decades have yielded substantial innovations in MR methods for the characterization of extracranial carotid atherosclerosis. Images obtained with these innovations under ideal conditions have given clinicians rich information about disease in the arterial wall and the hope for tools critically needed for adequate management of this insidious disease. Despite this, the great potential power of this technology has not made it into the routine clinical armamentarium. Indeed, because of the need for gadolinium-based contrast agents (GBCA), the long exam time (typically about 45 minutes to obtain the multiple contrasts in the 5 or 6 necessary sequences), and the steep learning curve required to interpret multi- contrast MRI most practitioners still revert to the simplified metric of diameter stenosis in assessing risk. After many collective years of investigations, the consortium of investigators collaborating on this proposal believes that the time is right to address these remaining limitations and ultimately shift the clinical paradigm. Overarching hypothesis: To achieve the great potential in the management of cervical carotid disease, a highly efficient and easily used MRI technique is required. Our hypothesis is that this can be accomplished using multi-parametric non-contrast MRI sequences coupled with the latest high signal to noise ratio (SNR) neck-shape-specific (NSS) RF coils and innovative machine learning (deep neural network) analysis methods. Aim 1: We will install identical RF coils, MRI sequences, and protocols at each of our 5 participating centers as well as rigorously test the accuracy of measurements and reproducibility of image quality from all centers. Aim 2: We will develop, train, and validate a user friendly, deep learning neural network system for the quantitative analysis of several key components considered to be present in the vulnerable atherosclerotic plaque. Aim 3: We will apply the analysis to a cohort of carotid disease subjects to establish the repeatability of the quantitative measures, as well as the accuracy of characterization in comparison to histopathology. Although we will develop and test the image quality, reproducibility and reliability in a network of highly skilled academic centers, we will design these methods to be applicable in the community hospital setting. At the conclusion of this project, we propose to have an integrated solution that can be used in subsequent investigations such as: the effect of pharmacologic intervention in modifying the composition of the plaque; studying the evolution of features of the untreated atheromatous disease over time; and, eventually, investigating the metrics that are predictive of deleterious outcomes, and that can be used in improving intervention strategies in this population. On successful completion, the RF coils and MRI sequences and analysis methods will be made available to other imaging centers in a manner that ultimately changes the paradigm of diagnosis and managing the treatment of cervical carotid atherosclerotic disease.