脑微出血定量检测中的正向磁化率计算方法研究

Cerebral microbleeds exist in 23% of Alzheimer’s disease, 40% of ischemic stroke, 68% of hemorrhagic stroke and 85% of vascular dementia. Accurate detection of cerebral microbleeds is important for the early diagnosis, treatment and mechanism study of cerebral diseases. Currently, most widely-used susceptibility weighted imaging method is non-quantitative. It is unable to provide comparison among results from different imaging conditions. Consequently, it is difficult to track the development of cerebral microbleeds if the data come from different MRI scanners. It also bring obstacle to the statistical analysis of large data from multiple MRI centers. Magnetic susceptibility, as an inherent physical property, may provide an objective criterion for microbleeds measurement. However, inverse-problem-based quantitative susceptibility mapping is limited by its ill-posed property. It leads to high sensitivity to noise and impose artifact on microbleeds detection. This project aims to propose a forward quantitative method for the measurement of cerebral microbleeds. It may improve the accuracy and stability of susceptibility measurement by using a linear regression method and a wrapping-independent phase calculating algorithm. The method is planned to be validated by phantom imaging at different MRI systems and using variable scanning parameters. Finally, the method is planned to be verified by clinical diagnosis of patients with cerebral microbleeds. The output of this project may provide a more accurate and objective method for the quantitative measurement of cerebral microbleeds and enable the comparison among different imaging conditions.

脑微出血存在于23%的阿尔茨海默症、40%的缺血性中风、68%的出血性中风和85%的血管性痴呆症中。脑微出血的准确检测有助于脑部疾病的早期预警，治疗方案的选择以及病理机制的研究。磁化率加权成像是目前应用最普遍的脑微出血检测方法，但它是非定量的，导致无法对不同成像条件下的结果进行对比，不利于对患者的病情跟踪和对大样本量病例的统计分析。磁化率作为物质的本质属性，可作为脑微出血的客观评价标准。但是以逆运算为基础的磁化率成像属于不适定性问题，其逆变换中的奇异值使结果对噪声敏感，并易产生干扰脑微出血检测的伪影。本课题拟提出一种正向磁化率计算方法以避免运算中的奇异性，采用线性回归的磁化率算法和原始卷绕数据下的相位算法以提高计算的准确性和客观性。本课题拟通过不同成像条件下的体模实验对方法的客观性进行评价，并在临床脑微出血患者的检测中进行验证。本课题将为脑微出血的临床磁共振检测提供一种客观准确的定量方法。

脑微出血是一种潜在的脑疾病隐患，与阿尔茨海默症、中风和血管性痴呆症相关，其早期的检测有助于脑部疾病的预防，治疗方案的选择以及病理机制的研究。磁共振成像中的磁化率加权像是目前应用最普遍的脑微出血检测方法，但它是非定量的，导致无法对不同成像条件下的结果进行对比，而且其假阳性率高，容易把正常组织错认为出血点。磁化率作为物质的本质属性，可作为脑微出血的客观评价标准。但是以逆运算为基础的磁化率成像属于不适定性问题，其逆变换中的奇异值使结果对噪声敏感，并易产生干扰脑微出血检测的伪影。本课题设计了一种基于相位梯度的磁化率成像方法，一方面能够直接处理包含卷绕的相位图像，另一方面又能够避免磁化率计算中逆问题所带来的奇异点伪影。该方法已经在数值模拟与体模实验中得到了定量的验证。在临床应用中，我们利用该方法计算获得了40例病人的脑部定量磁化率分布图。结果表明：低年资医生利用定量磁化率图判断脑微出血的结果要优于单独采用加权图像的判断结果，阳性预测值由55.4%上升到了71.0%，并且在不同的磁共振系统上有较好的一致性。