CAREER: Transforming Implantable Neural Interfaces through Computing: From Circuits to Systems

理解和设计大脑功能被认为是一项重大的科学挑战，因为它有可能给包括计算机和医学在内的许多领域带来革命性的变化。全植入式双向脑机接口(BBCI)是本项目的重点。这些是能够记录、处理和刺激神经活动的电子系统，它们在更好地理解大脑方面发挥着基础性作用。植入式BBCI将使神经科学家能够以前所未有的细节探索大脑功能，并帮助实现能够恢复残疾人行动能力、视力和大脑功能的神经假体设备。然而，BBCI面临的关键技术障碍正在阻碍神经科学的进步。现有的BBCI架构在功率或面积方面不能很好地扩展，以支持越来越多的记录和刺激通道，这是更精细地检查和控制大脑功能所需的。此外，神经刺激会产生伪像--大脑中的电干扰--这阻碍了进行神经记录的能力。最后，执行神经信号处理和与外部设备的数据通信所需的计算性能的期望水平消耗的功率超过可植入设备的热极限。该项目产生的技术将转化为一个完全可植入、生物兼容和多功能的闭环神经科学平台，以克服这些现有的挑战。在这个项目中，与神经科学家、医疗设备行业和制造合作伙伴的合作对实现这一目标至关重要。由此产生的平台将提供给更广泛的神经科学界，使实验能够达到前所未有的规模和范围，加快了解大脑的进展。该奖项通过研究和设计横跨数字/混合信号电路设计、体系结构、系统理论和系统集成的交叉技术，解决了BBCI在功率、面积、性能和记录质量方面的关键障碍。从算法划分到封装和电路设计，跨每个抽象级别利用特定于领域的结构是该项目的核心。这项工作分为三个方面：1)开发新的、计算增强的神经接口，以达到所需的效率和可扩展性水平；2)通过低能量计算探索与多输入多输出(MIMO)通信系统的域对应，这将使系统能够拒绝伪像并允许将记录集中到目标神经元集合；3)利用对神经信号处理算法的理解和在超低功率计算方面的初步结果，设计出在严重功率限制下满足BBCI处理要求的领域特定架构。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

Understanding and engineering brain function has been identified as a grand scientific challenge in recognition of its potential to revolutionize a number of fields including computing and medicine. Fully-implantable Bidirectional Brain Computer Interfaces (BBCI) are the focus of this project. These are electronic systems capable of recording, processing and stimulating neural activity, and they play a foundational role in enabling better understanding of the brain. Implantable BBCIs will enable neuro-scientists to explore brain function in unprecedented detail, and help realize neuro-prosthetic devices capable of restoring mobility, vision and brain function among the disabled. However, critical technological barriers facing BBCIs are hindering progress in neuroscience. Existing BBCI architectures do not scale well, in terms of power or area, to support ever-increasing numbers of recording and stimulation channels needed for finer examination and control of brain function. Furthermore, neural stimulation produces artifacts - electrical disturbances in the brain - that hamper the ability to perform neural recording. Finally, the desired level of computational performance required to perform neural signal processing and data-communication to external devices consumes power in excess of thermal limits of implantable devices. The technologies resulting from this project will be translated into a fully implantable, bio-compatible and versatile closed-loop neuroscience platform that overcomes these existing challenges. Collaborations with neuroscientists, the medical-device industry, and fabrication partners, to be pursued during this project, are critical to the realization of this goal. The resulting platform will be made available to the broader neuroscience community to enable experiments at unprecedented levels of scale and scope, accelerating progress toward understanding the brain. Both graduate and underrepresented minority students will be involved in the project.This award addresses the critical BBCI barriers of power, area, performance and recording quality by investigating and devising cross-cutting technologies that span digital/mixed-signal circuit design, architecture, systems theory and system integration. Exploiting domain-specific structure across every level of abstraction, from algorithm partitioning down to package- and circuit-design is central to this project. The effort is organized into three threads: 1) Development of novel, computationally-enhanced neural interfaces to achieve desired levels of efficiency and scalability; 2) Exploration of domain-correspondence to Multiple-Input Multiple Output (MIMO) communication systems through low-energy computing, which will allow systems capable of rejecting artifacts and allowing recording to be focused to a targeted set of neurons; and 3) Leveraging an understanding of neural signal processing algorithms and preliminary results in ultra-low power computing to devise domain specific architectures that meet BBCI processing requirements under severe power limitations.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.