Development and validation of automated analysis tools for clinical interpretation of vascular magnetic resonance images

人体的血管必须保持在健康的状态，才能使身体的其他部分正常运作。血管可能会出现两个主要问题：斑块堆积导致的堵塞(可能导致中风和心脏病发作)，以及血管壁僵硬，可能会通过多种原因导致过早死亡。我们可以使用磁共振成像(MRI)对血管以及这些血管中的阻塞进行成像。这种方法使用起来很安全，而且在大多数医院都可以使用。在牛津，我们已经开发了特定的血管成像方法，并研究了吸烟的影响，各种药物的影响，以及肥胖的影响。进一步的临床研究正在进行中。这些血管的磁共振成像的一个问题是，因为它们很小，所以很难准确地评估图像。如果我们寻找短期药物治疗后可能发生的微小变化，这尤其困难。这项研究开发了用计算机测量血管图像的方法。我们已经证明，半自动方法为测量血管壁厚提供了一种精确的工具。这项工作扩展了这种方法，通过心脏周期观察血管直径的动态变化，这可以用来告诉我们关于血管硬度的信息。这种方法在研究环境中的关键优势是它的精确度、速度和用户独立性。自动化程度更高的方法的另一个好处是，这将使医生从一项费时的工作中解脱出来。我们将开发的第二种方法是分析血管中斑块的核磁共振图像，以调查它是哪种类型的阻塞。这将形成一个单独的项目，将由一名DPhil(博士)学生在诺布尔教授和马修·罗布森博士的监督下进行。斑块可以包括三种不同的成分(脂肪中心、纤维帽和钙化组织)，每种成分对患者的健康都有不同的影响。通过采集3幅不同对比度的磁共振图像，我们可以区分这些成分。这种分析最好由计算机来执行，因为它需要同时考虑这3幅图像中的每一幅。其他小组已经展示了进行这种分析的方法，但这些方法不适用于一般的成像界，并且只应用于少量的患者图像。这项工作的一个重要部分是使我们开发的所有工具和技术可供所有临床研究人员使用。这项工作的初始目标是开发可以在研究环境中使用的方法，以便我们能够准确地监测疾病和药物治疗的效果。我们将在临床病例中验证这些方法，并使用不同类型的MRI扫描仪获取的数据来验证这些方法。我们将把该代码分发给该领域的关键研究小组(同时保护我们的知识产权)，以便进一步验证其临床用途并获得更广泛的反馈。随后，随着这些血管磁共振成像程序被所有医院广泛采用，我们预计这些小组也将使用我们将开发的分析方法。一旦这些方法被接受和验证，该项目的下一阶段(不是作为这笔赠款的一部分提供资金)将是将它们转化为临床产品，可以出售给普通临床社区。通过销售这样的产品，我们可以确保用户得到强有力的培训和支持。我们希望通过一家已有的临床心脏软件分销商销售这一产品。

The blood vessels of the human body must remain in a healthy condition to allow the rest of the body to function properly. Two major problems can occur in blood vessels; blockages due to the accumulation of plaque (which can cause stroke and heart-attack), and stiffening of the vessel wall that can lead to premature death through a number of causes. We can image the blood vessels, and blockages in these vessels using magnetic resonance imaging (MRI). This approach is safe to use, and is available in most hospitals. Here in Oxford we have already developed specific methods for vessel imaging, and have studied the effects of smoking, the effects of various drugs, and the effects of obesity. Further clinical studies are in the pipeline. One problem with MRI of these vessels is that, because they are small, it is quite difficult to evaluate the images accurately. It is especially difficult if we are looking for the small changes that might occur after a short course of drug treatment.This research develops ways of measuring the images of vessels by computer. We have already shown that semi-automated methods provide a precise tool for measuring the vessel wall thickness. This work extends that approach to look at the dynamic changes in the vessel diameter through the heart cycle, which can be used to tell us about vessel stiffness. The key strengths of this approach in the research environment are its precision, speed and user independence. One other benefit of a more automated approach is that is will relieve doctors of a laborious time consuming task.The second approach that we will develop is to analyse MRI images of plaque in vessels to investigate what type of blockage it is. This will form a separate project that will be taken on by a DPhil (PhD) student under the supervision of Professor Noble and Dr Matthew Robson. Plaque can include three different components (fatty centre, fibrous cap, and calcified tissue) each of these have different implications to the health of the patient. By acquiring 3 MR images, each with different contrast, we can distinguish each of these components. This analysis is best performed by computer as it needs to consider each of these 3 images at the same time. Other groups have already demonstrated methods for performing this analysis, but these methods are not available to the general imaging community, and have only been applied to small numbers of patient images. One important part of this work is to make all the tools and techniques that we develop available to all clinical researchers.The initial goal of this work is to develop methods that can be used in the research environment so that we can accurately monitor disease and the effects of drug treatments. We will validate these approaches on clinical cases, and validate these methods using data acquired on different types of MRI scanner. We will distribute this code to key research groups in the field (whilst protecting our intellectual property), in order to further validate its clinical utility and obtain extended feedback. Subsequently, as these vessel MR imaging procedures become widely adopted by all hospitals, we would expect these groups to also use the analysis methods that we will have developed. Once these approaches are accepted and proven, the next stage of the project (not funded as part of this grant) will be to convert them into a clinical product that can be sold to the general clinical community. By selling such a product we can ensure that the users get strong training and support. We would hope to sell this product via a pre-existing distributor of clinical cardiac software.