CAREER: Ultrasound-based Intent Modeling and Control Framework for Neurorehabilitation and Educating Children with Disabilities and High School Students

每年，美国有超过17000人经历脊髓损伤。患有不完全性脊髓损伤的人可能可以自主控制四肢，但与健全的人相比，他们的力量减弱了。神经康复是一种治疗方法，用于恢复肢体功能的个体遭受不完全性脊髓损伤；机器人或电刺激装置通过重复的辅助运动来促进人的神经系统的修复。该设备提供的辅助量是基于用户的剩余肌肉功能。然而，在治疗性运动中正确测量一个人的自愿努力是一项重大的技术挑战。测量的差异可能导致机器人提供太多或太少的帮助，这可能会减缓恢复并导致机器人辅助行走时摔倒。为了解决感知自愿努力的技术挑战，该项目将超声波成像与基于肌电图（即电信号）的踝关节肌肉测量相结合，以控制行走。超声成像的使用将允许直接可视化和测量肌肉活动，并最大限度地减少来自邻近肌肉的电信号干扰。超声成像也将研究作为优化电极放置的方法，以启动多平面踝关节运动。该研究将验证基于超声成像的自愿努力测量比单独基于肌电图的预测更准确的假设。通过以学生为主导，为有特殊需要的儿童构建超声成像和肌电图为基础的人机交互平台，将研究和教育结合起来。人机交互平台可以加速有特殊需要的儿童的学习，并为来自STEM中代表性不足群体的高中生提供研究机会。PI的长期职业目标是建立一个全尺寸的前馈肌肉骨骼模型，通过获取连接在不同肢体肌肉上的可穿戴超声波传感器的信号来预测人类的意图。为了实现这一目标，该项目将制定控制策略，使用超声成像来预测不完全性脊髓损伤患者的自主能力减弱，并在需要时使用混合外骨骼提供帮助。本课题的研究目标是：(1)建立基于超声成像的观测器来预测踝关节肌肉的自主力；(2)将预测结果应用于踝关节控制策略。将确定一个最佳的超声成像衍生替代信号来衡量自愿努力，并制定替代信号和自愿努力之间的功能映射。建立观测器和踝关节控制器的稳定性和收敛性条件。新的意图预测和控制框架将直接与基于肌电图的预测和控制方法进行比较。踝关节肌肉产生多平面踝关节运动的功能性电刺激也将被研究。针对预期的踝关节肌肉，将评估基于超声成像的反馈，以优化刺激电极的电流。该项目促进了一种突破性的康复治疗，使用无创、可穿戴的超声波传感器来收集、监测和控制不完全性脊髓损伤患者的肌肉活动。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Each year, more than 17,000 people in the United States experience a spinal cord injury. Individuals suffering from incomplete spinal cord injuries may have voluntary control of their limbs, but their strength is weakened compared to able-bodied persons. Neurorehabilitation is a therapeutic approach used to recover limb function in individuals suffering from incomplete spinal cord injury; robotic or electrical stimulation devices encourage repair of the person's nervous system through repeated, assisted exercise. The amount of assistance the device provides is based upon the user's remaining muscle function. However, correctly measuring a person's voluntary effort during therapeutic exercise is a significant technical challenge. Discrepancies in the measurement can cause the robot to provide too much or too little assistance, which can slow recovery and lead to falls during robot-assisted walking. To address the technical challenge of sensing of voluntary effort, this project will integrate ultrasound imaging with electromyography-based (i.e., electrical signals) measurement of ankle muscles that govern walking. The use of ultrasound imaging will allow direct visualization and measurement of muscle activity and minimizes interference of electrical signals from neighboring muscles. Ultrasound imaging will also be investigated as an approach for optimizing electrode placement to initiate multi-plane ankle movements. The study will test the hypothesis that ultrasound imaging-based measurement of voluntary effort is more accurate than electromyography-based prediction alone. Research and education are integrated through student-led construction of ultrasound imaging and electromyography-based human-machine interaction platforms for children with special needs. The human-machine interaction platforms may accelerate learning in children with special needs and offers research opportunities for high school-age students from underrepresented groups in STEM. The PI's long-term career goal is to build a full-scale, feedforward musculoskeletal model that predicts human intent by obtaining signals from wearable ultrasound sensors attached to different limb muscles. Toward this goal, the project will derive control strategies that use ultrasound imaging to predict weakened voluntary effort of a person with incomplete spinal cord injury and provide assistance as-needed during walking with a hybrid exoskeleton. The research objectives of the proposal are to: (1) formulate an ultrasound imaging-based observer to predict voluntary effort in ankle muscles and (2) use the prediction in an ankle control strategy. An optimal ultrasound imaging-derived surrogate signal to measure voluntary effort will be determined, and a functional mapping between the surrogate signals and voluntary effort will be formulated. The stability and convergence conditions for the observer and the ankle controller will be established. The new intent prediction and control framework will be directly compared to electromyography-based prediction and control methods. Functional electrical stimulation of the ankle muscles that produce multi-plane ankle movements will also be investigated. To target the intended ankle muscle, ultrasound imaging-based feedback will be evaluated to optimize current in the stimulation electrodes. The project facilitates a breakthrough rehabilitation therapy that uses non-invasive, wearable ultrasound sensors to collect, monitor, and control muscle activity of persons with incomplete spinal cord injury.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.