脑起搏器是一种可以治疗多种神经系统疾病和精神障碍疾病的有源植入式医疗设备。磁共振成像（Magnetic Resonance Imaging, MRI）是一种重要的医学成像手段，尤其适用于脑组织结构和脑神经活动成像。脑起搏器在MRI下实现全面应用对于脑疾病诊疗和脑科学研究都具有重要意义。脑起搏器在MRI下基本的安全性问题目前已得到初步解决，然而脑起搏器在MRI下产生的非预期刺激风险严重限制了脑起搏器在MRI下的进一步应用，而且非预期刺激风险尚未被充分证明不会引起安全问题。为此，本文围绕脑起搏器在MRI下的非预期刺激问题展开研究。首先，本文深入研究了脑深部刺激回路各部分的电学等效模型以及脑起搏器在MRI下感应电压的产生原理和特点，发现梯度感应电压是引起非预期刺激风险的主要成因，并且非预期刺激问题更可能发生于脑起搏器的电压刺激模式。基于上述结论，本文采用回路面积法对梯度感应电压的幅值范围进行估算，建立了非预期刺激的电学模型。参考标准ISO 10974（2018）设计了非预期刺激的评估方法和测试系统。采用噪声问题分析范式，建立了非预期刺激的噪声模型，提出了适用于抑制非预期刺激风险的阻抗调节方法和反馈补偿方法。其次，对基于阻抗调节方法的MRI兼容脑起搏器样机进行了详细设计。样机具有普通模式和MRI模式，普通模式下，样机按照设置参数输出脉冲；MRI模式下，样机通过内部电路提升刺激回路阻抗以降低梯度感应电压在组织两端形成的压降，与此同时，增大直流转换模块电压输出以保持组织两端正常的刺激幅值。测试结果表明，在常用治疗参数范围内，样机对于非预期刺激风险的抑制率在66%以上，在刺激幅值在不超过2.5 V的情况下，抑制率高达83%。最后，对基于反馈补偿方法的MRI兼容脑起搏器样机进行了详细设计。样机具有普通模式和MRI模式，普通模式下，样机按照设置参数输出脉冲；MRI模式下，样机采集刺激信号，计算采集信号与参考信号的误差，根据误差调节直流转换模块电压输出以保持组织两端正常的刺激幅值。测试结果表明，在常用治疗参数范围内，样机对于非预期刺激风险的抑制率约为74.5%。本文的工作为MRI兼容脑起搏器产品设计提供了方案框架和优化方向，对脑起搏器在MRI下的全面应用具有重要意义。

Deep brain stimulators are active implantable medical devices (AIMD) used to treat various neurological and mental disorders. Magnetic resonance imaging (MRI) is an important medical imaging method, particularly suitable for imaging brain tissue structure and neural activity. The comprehensive application of deep brain stimulators during MRI is of great significance for both the diagnosis and treatment of brain diseases and research in neuroscience. Although the basic safety issues of deep brain stimulators during MRI have been preliminarily addressed, the risk of unintended stimulation of deep brain stimulators during MRI severely limits their further application, and the safety of unintended stimulation has not been fully demonstrated. Therefore, this paper focuses on the unintended stimulation of deep brain stimulators during MRI.Firstly, this paper thoroughly investigates the electrical models of various parts of the deep brain stimulation (DBS) circuit and the principles and characteristics of induced voltage by these devices during MRI. It is found that gradient-induced voltage is the main cause of the risk of unintended stimulation, which is more likely to occur in the voltage stimulation mode of deep brain stimulators.Based on the above conclusions, this paper uses the circuit area method to estimate the range of gradient-induced voltage and establishes an electrical model of unintended stimulation. Referring to ISO 10974 (2018), an evaluation method and test system for unintended stimulation are designed. By adopting the paradigm of noise problem analysis, a noise model of unintended stimulation is established, and impedance regulation methods and feedback compensation methods suitable for suppressing the risk of unintended stimulation are proposed.Secondly, a detailed design of an MRI-compatible deep brain stimulator prototype based on impedance regulation method is conducted. The prototype has two modes: normal mode and MRI mode. During the normal mode, the prototype outputs pulses according to the parameter setting; during the MRI mode, the prototype increases the impedance of the stimulation circuit through internal circuits to reduce the voltage drop formed at both ends of the tissue due to gradient-induced voltage, while increasing the voltage output of the DC conversion module to maintain normal stimulation amplitude at both ends of the tissue. The test results show that within the common therapeutic parameter range, the suppression rate of unintended stimulation risk is above 66%, and it reaches as high as 83% when the stimulation amplitude does not exceed 2.5 V.Finally, a detailed design of an MRI-compatible deep brain stimulator prototype based on feedback compensation method is conducted. The prototype also has two modes: normal mode and MRI mode. During the normal mode, the prototype outputs pulses according to the parameter setting; during the MRI mode, the prototype collects stimulation signals, calculates the error between the collected signal and the reference signal, and adjusts the voltage output of the DC conversion module based on the error to maintain normal stimulation amplitude at both ends of the tissue. The test results show that within the common therapeutic parameter range, the suppression rate of unintended stimulation risk is approximately 74.5%.The paper provides a framework and optimization direction for the design of MRI-compatible deep brain stimulator products, which is of great significance for the comprehensive application of deep brain stimulators during MRI.