Fully automatic mapping of cerebral perfusion territories using Magnetic Resonance Imaging

该项目旨在完成一种加速脑灌注区域测绘方法的开发，而无需手动规划单个采集。使用动脉自旋标记可以在不使用任何外部造影剂的情况下进行灌注成像，从而使该方法对反复采集和健康志愿者的应用具有吸引力。该方法的改进允许全脑灌注成像（非选择性），主要动脉（颈动脉和椎动脉）提供的区域以及较小的颅内动脉。然而，选择性收购的一个主要缺点是需要在收购之前进行耗时的规划程序。当然，需要加快这一过程的方法。这些方法是存在的，但由于MRI的物理原理，它们只允许获得相对而不是定量的灰度值，因此受到限制。因此，需要新颖的算法，它足够严格，只能检测感兴趣的结构，但又足够灵活，以适应个体患者的解剖结构。本项目的目标不仅是算法优化，而是使用组合方法优化MR序列，以最大限度地提高感兴趣组织的对比度，同时改进检测方法的可靠性。在这个项目中要处理的另一个主要限制是选择性动脉自旋标记方法的扫描持续时间较长。使用区域方法，通常图像采集时间随着获取的区域数量线性增加。这增加了不必要的整体扫描持续时间，并可能导致扫描期间更明显的运动伪影。这在选择性绘制颅内小动脉时尤为重要。因此，本项目进一步考虑了在不增加扫描时间的情况下，在单次扫描中获取单个区域的方法。总之，一种自动计划选择性灌注采集并减少整体扫描时间的方法将是本项目的成果。

This projects aims towards the complete development of an accelerated brain perfusion territory mapping method without the need for manually planning the individual acquisitions. Using Arterial Spin Labeling it is possible to perform perfusion imaging without the application of any external contrast material, thereby making this method attractive for recurrent acquisitions as well as for the application in healthy volunteers. Modifications of the method allow for imaging of whole brain perfusion (non-selective), territories supplied by the major arteries (carotid and vertebral arteries) as well as smaller, intracranial arteries. A major disadvantage of selective acquisitions is however the need for a time-consuming planning procedure prior to the acquisition. Naturally, methods to speed up this process are desired. These approaches exist, yet are limited due to the physical principles of MRI, which allow only for obtaining relative rather than quantitative gray values. Therefore, novel algorithms are needed, which are stringent enough to only detect the structures of interest, yet are flexible enough to adapt to individual patients anatomies. This projects aims not only for algorithm optimization, but uses a combined approach to optimize MR sequences to maximize the contrast of the tissues of interest, while the detection methods are improved regarding their reliability. Another major limitation to be dealt with during this project is the longer scan duration of selective Arterial Spin Labeling approaches. Using territorial methods, generally the image acquisition time increases linearly with the number of acquired territories. This increases the overall scan duration unnecessarily and potentially leads to more pronounced motion artifacts during scanning. This is especially critical in mapping the smaller intracranial arteries selectively. Therefore, a method to acquire the individual territories within a single scan without increasing the scan time is furthermore considered in this project. In conclusion, a method to automatically plan the selective perfusion acquisitions in combination with a reduced overall scan time will be the outcome of this project.