An Enhanced Transcranial Magnetic Stimulator for Advancing Clinical Neuroscience and Minimally-Invasive Brain Therapies

经颅磁刺激（TMS）是一种非侵入性的脑回路刺激技术。经颅磁刺激是神经科学的一种研究工具，在临床治疗中有很好的转化机会。目前经颅磁刺激的临床应用包括重度抑郁症、强迫症和偏头痛。正在探索的经颅磁刺激疗法包括中风后康复、中枢性疼痛、成瘾和耳鸣。然而，经颅磁刺激疗法有待改进。例如，在抑郁症中，只有大约1/3的患者得到缓解。为了进一步利用TMS，该项目将探索两个互补的技术领域。首先，我们认为扩大增产参数空间是有益的。从历史上看，大多数TMS应用使用1 - 20hz的频率，连续或简单的重复脉冲。然而，新的范例正在出现；“刺激疗法”最近显示出比连续经颅刺激疗法更短的治疗间隔有望治疗抑郁症（Blumberger, Lancet, 2018）。此外，其他神经调节疗法，如深部脑刺激，使用明显更高的频率来达到效果。脉冲形状是差分参与神经基质的另一个自由度，但尚未得到充分探索，因为系统通常在单一共振频率下运行。这些数据表明，具有扩展参数能力的经颅磁刺激可能开启新的治疗用途或改善结果。其次，为了充分利用这些扩展功能，我们认为研究人员需要一种系统地搜索参数空间的方法。即使有了今天的选择，只有一小部分集群在实践中使用，这些设置的机制尚不清楚（Klimjai, Annals Phys Rehab Med, 2015）。与θ波爆发类似，其他模式可能会出现，提供额外的好处，但探索空间的途径必须是可处理的。最后，颅磁刺激兼容的脑电图系统是可用的，但没有明确的算法来衡量颅磁刺激潜在的有益或负面影响。本论文将设计并测试一个增强型TMS系统，以解决这些缺点。首先，该系统将提供具有扩展参数集的TMS，包括更高的速率，扩展的模式能力，以及可调谐电路的脉冲整形变化。该项目的这一部分将需要系统级集成电子、磁线圈和神经激活模型。论文的第二部分是设计一种用于搜索刺激参数空间的强化学习算法。生理传感器将用于实时评估受试者的大脑状态。然后应用刺激的扰动，并使用对大脑状态的影响来调整下一个参数运行。使用强化（机器）学习的方法，我们将探索基于模型和“黑盒”的TMS参数空间探索，以寻找更优的刺激范式。为了证明新研究系统的实用性，论文将包括对选定神经元网络的长期增强和抑制的测量。我们将使用已建立的方法来调制运动和语言中心，作为方法验证的客观标记。最终成果将是一个概念验证仪器和参数优化框架，用于推进经颅磁刺激临床神经科学研究和应用。商业合作伙伴：Magstim， Whitland, Wales

Transcranial Magnetic Stimulation (TMS) is a non-invasive technology to stimulate brain circuits. TMS is a research tool for neuroscience, with promising translation opportunities for clinical treatments. Current TMS clinical applications include major depressive disorder, OCD and migraine. TMS therapies under exploration include post - stroke rehabilitation, central pain, addictions and Tinnitus. However, TMS therapy needs to be improved. For example, in depression, only about 1/3 of patients achieve remission. To further TMS utility, this project will explore complementary two areas of technology. First, we believe expanding the stimulation parameter space is beneficial. Historically, most TMS applications use frequencies of 1 - 20 Hz, continuously or in simple repetitive bursts. Yet new paradigms are emerging; "Thetaburst Stimulation" recently showed promise to treat depression with shorter treatment intervals than continuous TMS (Blumberger, Lancet, 2018). In addition, alternative neuromodulation therapies like Deep Brain Stimulation use significantly higher frequencies for effect. Pulse shape is another degree of freedom for differentially engaging neural substrates, but is not fully explored, as systems generally operate at a single resonance frequency. These data suggest TMS with expanded parameter capabilities may unlock novel therapeutic uses or improve outcomes. Second, to make the most of these extended capabilities, we believe researchers need a methodology for systematically searching the parameter space. Even with today's options, only a small set of clusters are used in practice, and the mechanisms of these settings are not clearly understood (Klimjai, Annals Phys Rehab Med, 2015). Similar to theta burst, other patterns might emerge that provide additional benefit, but the pathway for exploring the space must be tractable. Finally, TMS compatible EEG systems are available but no clear algorithms to measure potentially beneficial or negative effects of TMS have been validated.This thesis project will design and test an enhanced TMS system that addresses these shortcomings. First, the system will deliver TMS with an expanded parameter set, including higher rates, extended pattern capability, and variations in pulse shaping with tunable circuits. This portion of the project will require systems-level integration of electronics, magnetic coils, and neural activation models. The second component of the thesis will be to design a reinforcement learning algorithm for searching the stimulation parameter space. Physiological sensors will be used to estimate the subject's brain state in real-time. Perturbations of stimulation will then be applied, and the effect on brain state used to adjust the next parameter run. Using methods from reinforcement (machine) learning, we will explore both model-based and "black box" explorations of the TMS parameter space to in search of more optimal stimulation paradigms.To demonstrate the utility of the new research system, the thesis will include measurements of Long Term Potentiation and Inhibition of selected neuronal networks. We will use established methods for modulation of motor and speech centers as objective markers for method validation. The final deliverable will be a proof-of-concept instrument and parameter optimization framework for advancing TMS clinical neuroscience research and applications.Commercial partner: Magstim, Whitland, Wales