高级ASL Perfusion MRI技术研究

Cerebral blood flow (CBF) is an important physiological measure. Because CBF reflects regional brain metabolism, it has long been assessed as a marker for regional brain function in basic neuroscience and clinical studies. Arterial spin labeling (ASL) perfusion MRI is a noninvasive technique for quantifying CBF and has been increasingly used in various basic neuroscience research and clinical studies. Limited by the T1 decay of the labeled spins, ASL MRI has a low signal-to-noise-ratio (SNR), which has ever been the major force driving the whole ASL MRI technique development. Prior technique developments have been focused on either more advanced ASL MRI sequences or better signal processing. A combination of them should provide a more optimal solution for high-quality ASL CBF quantification, as planned in this project. The first aim of this study is to further optimize and develop the 3D fast-spin-echo spiral readout background suppressed pseudo-continuous ASL sequence, a state-of-art that we have recently developed, to further increase the acquisition speed so as to reduce the blurring effect or increase spatial resolution. The second aim is to develop more effective ASL MRI denoising methods. Specific experiments are designed for method validations. Using the proposed post-processing methods, we will identify an optimal minimum scan time as guidance for routine ASL MRI applications. With completion of this project, we will disseminate the ASL sequence and post-processing methods as free packages for academic users.

脑灌注(cerebral blood flow又称脑血流)是非常重要的生理指标。ASL perfusion MRI是无创可多次重复使用的脑血流测量技术。在认知科学和脑疾病研究中有广泛的应用前景。由于标记的信号量有限，ASL MRI获取的灌注信号噪声偏大，迫切需要研发新的技术以提高其信噪比。提高ASL图像信噪比可以通过新的成像技术或新的数据处理方法来实现。以往的研究关注的都是其中的一个问题，更优化的解决方案是合二为一。这也是本项目的立项基础和根本目的。本项目第一个目标是开发性能更佳的ASL MRI成像技术，第二个目标是研发更好的ASL信号处理方法以提高ASL MRI成像质量和血流测量精度。本项目是建立在申请人以往12年的ASL工作基础之上，研究目标也都有充分的预研数据支持。项目的成功实施将会使得基于ASL MRI的脑血流测量更为快速准确和可靠，进一步推动ASL在临床和认知科学研究中的应用。

提出并发表了基于图层的ASL MRI伪迹点去除算法。该算法相比以往的基于整个三维图像的算法而言极大地改进了局部图像质量和有效地保留了有用的图像信息，避免了以往算法中鱼沙俱下的做法。开发了基于低秩和稀疏分解的ASL 数据去噪方法，以往去噪主要关注空间方向，这个方法同时还关注了时间方向。该方法能有效去除大部分噪声，将信噪比提高15%以上。首次提出并发表了不需要字典的磁共振指纹快速匹配技术，比现有办法快1000多倍。以往的算法都是地毯式搜索，耗时耗力，如果目标不在收缩范围之内时就不可能得到正确答案。新的算法提供了一个理论框架，确保了快速有效的搜索过程，实现了收缩范围自动调节，搜索精度自动调节。