Data-driven Modeling and Ultrasound-based Control of Afferent Nerve Stimulation for Tremor Suppression

项目摘要 在美国，超过1100万人受到无意和无法控制的节奏运动的影响。 由于帕金森震颤、特发性震颤和小脑震颤。手部出现震颤的患者 和手臂面临日常生活活动的困难。电刺激，通过刺激电机工作 拮抗肌的神经是当药物无效时抑制震颤的潜在可穿戴选择， 但在追求有效但侵入性（和昂贵）的脑手术之前。然而， 运动神经刺激是由于高刺激强度引起的刺激引起的肌肉疲劳和不适。一个 一种有趣的新刺激方法靶向传入神经纤维，以抑制拮抗肌的震颤。这些 纤维将关于肌肉位置和速度的感觉信息传递回脊髓。通过刺激这些纤维， 可以调节脊髓神经回路，其又由于下行震颤输入而抑制肌肉活动 从大脑中。新方法使用低刺激强度，这是舒适和抗疲劳。然而，在这方面， 调整其刺激参数以持续破坏震颤仍然具有挑战性，这主要是由于 在受刺激的传入神经和下行震颤神经之间可能发生许多相互作用， 输入。这些相互作用通过复杂的神经回路发生，其建模是困难的和计算的 用于实时确定刺激参数。肌肉速度和长度的直接测量 超声波的变化可以帮助创建传入刺激的数据驱动模型，并帮助设计个人 特定的传入刺激参数。然而，超声波从未用于震颤抑制控制。 将超声衍生的肌肉活动映射到振荡肢体位移的实时算法和模型是 尚未确定。需要这些算法和模型来自动化个体特异性刺激参数 用于抑制震颤。最后，对于未来的临床翻译， 肌肉需要发展。以下两个具体目标将涉及这项工作。SA 1：对传入神经建模 反馈在震颤活动和设计和验证模型通知传入刺激策略。 SA2：开发可穿戴超声阵列。如果成功，传入刺激的数据驱动模型将 自动化个性化的震颤抑制干预。数据驱动建模的未来临床翻译， 传入刺激技术，沿着可穿戴超声阵列，将通过协助改善生活质量 抑制明显的远端震颤

PROJECT SUMMARY Over 11 million people in the United States are affected by unintentional and uncontrollable rhythmic movements due to parkinsonian tremor, essential tremor, and cerebellar tremor. Individuals experiencing tremors in the hands and arms face difficulty performing activities of daily living. Electrical stimulation that works by stimulating motor nerves of antagonistic muscles is a potential wearable option for tremor suppression when medication is ineffective, but prior to pursuit of effective yet invasive (and costly) brain surgery. However, the two significant drawbacks of motor nerve stimulation are stimulation-induced muscle fatigue and discomfort due to high stimulation intensity. An intriguing new method of stimulation targets afferent nerve fibers to inhibit tremor in antagonistic muscles. These fibers relay sensory information on muscle position and velocity back to the spinal cord. By stimulating these fibers, spinal neural circuitry can be modulated that in turn inhibits muscle activity due to the descending tremor inputs from the brain. The new method uses low stimulation intensity, which is comfortable and fatigue resistant. However, modulating its stimulation parameters to continually disrupt tremors remains challenging, largely due to the numerous interactions that can occur between stimulated afferent nerves and the descending tremorgenic neural inputs. These interactions occur through a complex neural circuitry whose modeling is difficult and computationally intensive for determining stimulation parameters in real-time. Direct measurements of muscle velocity and length changes with ultrasound can help create a data-driven model of afferent stimulation and help design individual- specific afferent stimulation parameters. However, ultrasound has never been used for tremor suppression control. Real-time algorithms and models that map ultrasound-derived muscle activity to oscillating limb displacement are yet unestablished. These algorithms and models are needed to automate individual-specific stimulation parameters for tremor suppression. Lastly, for future clinical translation wearable ultrasound arrays that monitor multiple muscles need to be developed. The following two specific aims will address this work. SA 1: To model afferent feedback during tremor activity and design and validate a model-informed afferent stimulation strategy. SA2: To develop a wearable ultrasound array. If successful, a data-driven model of afferent stimulation will automate an individualized tremor suppression intervention. A future clinical translation of data-driven modeling, afferent stimulation technology, along with the wearable ultrasound array will improve the quality of life by assisting in suppressing prominent distal tremors.