基于去卷积方法的MRI灌注成像平均渡越时间估计的研究

发布时间：2018-12-13 22:04

【摘要】：磁共振灌注成像是一种功能磁共振成像技术,它可以反映组织的微血管分布和血流灌注情况,能够提供解剖结构、血管成像和组织功能状态的多种血流动力学方面的信息。在磁共振灌注成像动力学参数中,平均渡越时间(MTT)对区分正常脑和缺血脑非常敏感,是诊断脑缺血疾病非常重要的参数。本论文首先对磁共振灌注成像技术的基本原理以及在脑部疾病和其他组织疾病中的应用进行了综述。进而,对MRI脑血灌注成像的基本原理进行公式推导,得出通过去卷积估算平均渡越时间的方法,并对其理想的动力函数进行了仿真模拟。而直接去卷积方法计算量比较大且繁琐,因此本文采用了基于加权广义交叉验证(W-GCV)的Tikhonov正则化法、总体最小二乘正则化法和修正截断奇异值正则化法进行去卷积,对平均渡越时间求解,并对驻留函数进行仿真分析。以及对三种正则化法的抗噪性能进行评估和受到示踪剂延迟的影响进行分析讨论,结果表明,随着信噪比的增大,信噪比对三种正则化的估算结果影响逐渐减小。基于W-GCV的Tikhonov正则化方法所得平均渡越时间与参考值更接近,偏差程度最小;随着示踪剂延迟的增加,三种正则化方法估算出的平均渡越时间也在增加,且增加的值小于延迟的时间。最后,利用基于W-GCV的Tikhonov正则化方法对真实脑灌注图像进行了动力参数伪彩图成像,并选取正则化方法所得的估计结果,结合脑血流量(CBF)伪彩图,分析了大脑感兴趣区域的缺血情况。结果表明,患侧与健侧的平均渡越时间比值大于1.63,说明大脑处于缺血状态。且从脑血流量伪彩图可以得出患侧的CBF值在20~35ml/min之间,提示N蛋白合成停止,若血量不继续下降脑组织仍然存活。以上结果可为脑缺血的临床诊断提供依据。
[Abstract]:Magnetic resonance perfusion imaging is a kind of functional magnetic resonance imaging technology, which can reflect the microvascular distribution and blood perfusion of tissues, and can provide various hemodynamic information of anatomical structure, vascular imaging and tissue functional state. Among the dynamic parameters of magnetic resonance perfusion imaging, the mean transit time (MTT) is very sensitive to differentiating the normal brain from the ischemic brain, and it is an important parameter in the diagnosis of cerebral ischemic diseases. In this paper, the basic principle of magnetic resonance perfusion imaging and its application in brain and other tissue diseases are reviewed. Then, the basic principle of MRI cerebral blood perfusion imaging is deduced, and the method of estimating the average transit time by deconvolution is obtained, and its ideal dynamic function is simulated. However, the direct deconvolution method is more complicated and complicated, so the Tikhonov regularization method based on weighted generalized cross validation (W-GCV), the total least square regularization method and the modified truncated singular value regularization method are used to deconvolution. The average transit time is solved and the resident function is simulated. The anti-noise performance of the three regularization methods is evaluated and the influence of tracer delay on the performance of the three regularization methods is analyzed and discussed. The results show that with the increase of SNR the influence of SNR on the estimation of the three regularization results decreases gradually. The average transit time obtained by Tikhonov regularization method based on W-GCV is closer to the reference value and the deviation is minimum. With the increase of tracer delay, the average transit time estimated by the three regularization methods is also increasing, and the increase is less than the delay time. Finally, the real cerebral perfusion image is imaged by Tikhonov regularization method based on W-GCV, and the estimated results of the regularization method are selected, combined with the (CBF) pseudocolor image of cerebral blood flow. Ischemia in the region of interest to the brain was analyzed. The results showed that the average transit time ratio between the affected side and the healthy side was greater than 1.63, indicating that the brain was in ischemic state. Moreover, the CBF value of the affected side was between 20~35ml/min and the pseudo-color map of cerebral blood flow, suggesting that the synthesis of N protein was stopped, and if the blood volume did not continue to decline, the brain tissue still survived. The above results may provide the basis for the clinical diagnosis of cerebral ischemia.
【学位授予单位】：河北工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP391.41;R445.2

【参考文献】