脑深部电刺激（Deep Brain Stimulation，DBS）又称脑起搏器，是治疗多种脑疾病的有效疗法和开展脑科学研究的强大工具。磁共振成像（Magnetic Resonance Imaging，MRI）是临床最重要的成像手段之一，DBS与MRI结合具有重大的临床价值和科研意义。然而DBS在MRI下存在诸多兼容性问题，其中电极的射频（Radio Frequency，RF）致热是最主要因素。由于直接威胁着患者的健康安全，电极RF致热风险严重限制了DBS在MRI下的应用推广。电极RF致热过程受到多方面因素影响，规律复杂，而且个体差异大，因此在MRI下监测电极温升尤为重要。基于质子共振频率（Proton Resonance Frequency，PRF）的磁共振测温法是MRI下重要的无创温升监测手段，应用前景广阔。但该方法面临电极干扰PRF测温信号、测温与常规扫查序列时间不同步以及如何可靠衡量RF致热风险等问题，为温升监测和安全性评估带来巨大挑战。为此，本文主要做出了以下工作：首先，电极引发的静磁场和射频场干扰是造成MRI伪影的重要因素，本文基于对电极周围的PRF测温信号的模拟和重建，发现电极引发的静磁场干扰是影响PRF信号的主要因素。进一步地，本文分析并验证了伪影下PRF信号的统计模型，利用统计特性，有效识别出了电极周围可靠的测温区域。其次，本文构建了能够克服伪影干扰，反求电极表面温升的解决方案。利用PRF信号的统计模型和模拟传热过程，将温升反求问题转化为模型参数估计问题，提出了基于极大似然估计的电极温升反求和不确定度评估方法。与传统的直接拟合相比，本文提出的新方法能使温升反求误差降低50 %以上。最后，针对电极RF致热在体监测的实际应用场景，本文提出了射频致热的小温升检测方法，能够在热损伤出现前检测出电极温升，提升了RF致热风险评估的可靠性。进一步地，提出了基于降温段测温数据的温升反求，解决了测温与扫查序列不同步问题。进而，利用动物试验证明了上述方法在活体内的可行性。最终成功应用于3.0 T磁共振兼容脑起搏器临床试验，保障了患者安全。本文的研究工作为DBS在MRI下的应用提供了安全反馈和切实保障，同时也为其他植入物在MRI下的RF致热监测提供了理论和方法创新。
Deep Brain Stimulation (DBS) is an effective therapy for the treatment of various brain diseases and a powerful tool for brain research. Magnetic resonance imaging (MRI) is one of the most important clinical imaging methods. The combination of DBS and MRI has great clinical value and scientific significance. However, DBS devices have many compatibility problems under MRI, among which the Radio Frequency (RF) heating of the lead is the most important one. Because of the direct threat to the patients’ safety, the risk of RF heating severely limits the application and generalization of DBS under MRI. The RF heating of the lead is affected by various factors, their effects are not only complicated, but also vary greatly among individuals. Thus, it is particularly important to monitor the lead heating under MRI. Magnetic resonance thermometry based on proton resonance frequency (PRF) is an important non-invasive method for temperature monitoring and has broad application prospect. However, it faces various problems such as the interference to PRF signal from the lead, asynchronization between the thermometry and normal scanning sequence, and how to reliably measure the risk of RF heating. These problems bring huge challenges to monitoring the lead heating and safety assessment. Based on these issues, this thesis conducted the following studies.First, as the static magnetic field and RF field interference induced by the lead are important factors that cause artifacts under MRI, the PRF signal around the lead was simulated and reconstructed to demonstrate their influence. It was found that the static magnetic field interference caused by the lead was the main factor affecting the PRF signal. Furthermore, the statistical model of the PRF signal under artifacts was analyzed and verified. Based on the proposed statistical properties, a region for reliable PRF thermometry around the lead was identified.Second, a solution for the inverse calculation of the lead heating was established which could overcome the interference of the lead artifacts. Using the statistical model of the PRF signal and the thermal stimulation, the inverse problem for heating calculation was transformed into a problem of model parameter estimation. A method based on the maximum likelihood estimation was proposed together with a technique for uncertainty assessment, which were used to solve the inverse calculation of lead heating. Compared with the traditional direct fitting, the proposed new method could reduce the error of heating calculation by more than 50%.Third, for the application scenarios of the lead heating monitoring, a small heating detection method was proposed, which could detect the lead heating before thermal damage occurs, improving the reliability of RF heating risk assessment. Furthermore, the inverse calculation of lead heating based on PRF signal acquired during cooling section was established to solve the asynchronization between the thermometry and normal scanning sequence. Animal study was used to prove the in vivo feasibility of the above method. Finally, it was successfully applied to the clinical trial of 3.0T MRI compatible DBS to ensure patients’ safety.This thesis provides safety feedback and practical guarantee for the application of DBS in MRI, and also provides theoretical and methodological innovations for monitoring the RF heating of other implants under MRI.
