CAREER: Next-Generation Neural-Machine Interfaces for Electromyography-Controlled Neurorehabilitation

全世界数百万患者的生活受到上肢丧失或损伤的严重影响。一项新兴技术，基于肌电图（EMG）的神经机器接口（NMI），通过神经义肢为包括截肢者、中风幸存者和脑瘫患者在内的这类人群提供了巨大的功能恢复潜力。该技术感知来自肌肉的生物电信号，对其进行解读以识别患者的预期动作，并做出决定来控制神经康复应用（例如，假肢）。虽然神经康复系统的设计在过去几十年里取得了显著进展，但目前还没有一种系统能够满足商业和临床实施所需的所有技术规范。该项目采用计算机工程方法来改进基于肌电图的NMI技术的功能和鲁棒性。将开发用于管理传感器状态和实时响应的软件，并实施新的计算平台来处理响应性神经康复应用所需的大规模数据密集型计算。该项目通过几个途径将研究和教育结合起来：用嵌入式研究经验加强本科课程，开发一个关于神经机器接口的大规模开放在线课程，以及通过社区大学开展K-12的外展计划。PI的长期职业目标是开发连接人类的下一代nmi，并在神经康复研究中探索大数据和深度学习技术。为了实现这一目标，该项目的目标是：(1)开发新的硬件和软件方法，以便在基于肌电图的实时NMI中使用高密度网格传感技术，以提高NMI的功能和鲁棒性；(2)开发新的计算技术，使计算能力和存储容量不再是NMI神经康复研究进步的障碍。PI将首先解决将高密度肌电网格应用于实时nmi的挑战，通过使用网格状态感知和响应引擎来密切监测EMD网格的状态并做出相应的响应。高密度肌电图网格带来的计算负担问题将通过神经形态计算系统的发展得到解决。最后，将开发一个分层计算平台，提供足够的计算和存储能力，以实现实时响应和可移植性。基于肌电图的NMI设计的见解和进步将显著提高肌电图控制的神经康复系统的可靠性和功能。此外，所开发的NMI方法和工具也适用于神经康复应用以外的研究领域，如脑机接口。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。

The lives of millions of patients worldwide are severely impacted by upper extremity loss or impairment. An emerging technology, electromyography (EMG)-based Neural Machine Interface (NMI), offers enormous potential in the restoration of function through neuroprosthetics for this population, including amputees, stroke survivors, and cerebral palsy patients. The technology senses bioelectrical signals from muscles, interprets them to identify the intended movement of the patient, and makes decisions to control neurorehabilitation applications (e.g., a prosthetic limb). While neurorehabilitation system design has progressed remarkably over several decades, no system is currently capable of meeting all desired technical specifications for commercial and clinical implementation. This project takes a computer engineering approach toward improving EMG-based NMI technology functionality and robustness. Software will be developed for managing the sensor status and real-time responses, and novel computing platforms will be implemented to handle the large-scale, data-intensive computations required for responsive neurorehabilitation applications. The project integrates research and education through several avenues: enhancement of undergraduate curricula with embedded research experiences, development of a massive open online course on neural machine interface, and initiation of a K-12 through community college outreach program. The PI's long-term career goal is to develop next-generation NMIs that will connect people and enable the exploration of big data and deep learning technologies in neurorehabilitation research. Toward this goal, the project's objectives are to (1) develop new hardware and software methods to enable the use of high-density grid sensing technology in real-time EMG-based NMIs to improve the functionality and robustness of the NMIs and (2) develop new computing technologies so that computing power and storage capacity are no longer barriers to the advancement of NMI neurorehabilitation research. The PI will first address the challenge of applying high-density EMG grids to real-time NMIs by employing a Grid Status Awareness and Response Engine to closely monitor the status of the EMD grids and respond accordingly. The issue of computational burden posed by the high-density EMG grids will then be tackled through the development of a neuromorphic computing system. Finally, a hierarchical computing platform that provides sufficient computational and storage capabilities to enable real-time response and portability will be developed. Insights and advancements made here in EMG-based NMI design will markedly improve reliability and functionality of EMG-controlled neurorehabilitation systems. Additionally, the developed NMI methods and tools are applicable to research fields beyond neurorehabilitation applications, such as brain-computer interfaces.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.