Micro-Electronics for autonomous neural implants

可植入的神经接口可用于将人脑连接到人工电子电路，从而允许例如通过思想或治疗各种损伤和疾病来控制计算机。一个这样的例子是，通过绕过受损的神经连接并允许控制假肢来治疗脊髓损伤的可能性。这是通过在神经组织中插入电极来实现的，这些神经组织连接到仪器电路，记录活跃神经元产生的电势并解码它们的含义。虽然过去的解决方案通常依赖于记录单个神经元产生的细胞外动作电位(EAP)，也称为神经棘波，但该项目的目标是专注于局部场电位(LFP)的获取和处理。这是因为记录EAP的植入物通常由于宿主的异物反应而受到寿命限制的阻碍，导致疤痕组织生长充当空间和频率低通过滤器，从而限制了记录的高频EAP的保真度。LFP的低频特性可以显著降低疤痕组织生长对记录的影响。一些涉及到的挑战包括电极材料的选择。这必须确保LFP在与脑脊液接触时在频段内增加的热噪声最小，同时具有化学惰性和医疗无害。初步结果表明，Nb(Nb)是一种很有前途的材料，适合于这种记录，因为它是已知的生物惰性材料，通常用于例如牙科植入物。它的极化性导致LFP频段的噪声功率密度通常比常用的铂和钨更小，使其成为适合神经记录的候选材料。这将通过电解液中的直接噪声测量进行研究，并随后通过直接体内测量进行验证。另一个挑战在于采集电子设备的发展，它受到其功率消耗的限制，这些限制由神经组织中80 mW/cm^2的安全散热限制所支配。允许减少使用能量的技术之一是无时钟，也被称为连续时间(CT)，信号采集。这种方法导致了与活动相关的电路，只有在其输入端检测到活动时才使用能量。该项目的目的之一是调查这种电路对于获取LFP的适宜性和可能的优势。由于这种采集和采样过程的性质在很大程度上仍然未知，因此使用数学模拟来研究它们。此外，还有集成电路的设计和验证理论模型假设的实际测量。这已经导致了为防止混叠而需要满足的最低要求的表述，类似于适用于普通抽样的奈奎斯特定理。此外，还提出了一种减小CT采样中闪烁噪声的新方法，并从理论上进行了验证。这项工作是ENGINI项目的一部分，目的是设计下一代神经植入物，通过完全无线、尺寸最小、靶向LFP记录和允许形成分布式植入物网络来实现卓越的慢性化。该研究与EPSRC的以下研究领域一致：辅助技术、康复和肌肉骨骼生物力学；微电子设备技术

Implantable neural interfaces can be used to connect human brains to artificial electronic circuits allowing e.g. control of computers by thoughts or treatment of various injuries and illnesses. An example of such is the possibility of treating spinal cord injuries by bypassing the damaged neural connection and allowing control of artificial limbs. This is achieved by inserting electrodes into neural tissue connected to instrumentation circuits recording electronic potentials generated by active neurons and decoding their meaning. While past solutions typically relied on recording of extracellular action potentials (EAPs), also known as neural spikes, generated by single neurons, the aim of this project is to focus on acquisition and processing of local field potentials (LFPs). This is as EAP-recording implants are typically hindered by limited lifetime due to the host body's foreign body response leading to scar tissue growth acting as a spatial and frequential low-pass filter, hence limiting the fidelity of recorded high-frequency EAPs. The low-frequency nature of LFPs allows for a significant reduction of the effect scar-tissue growth has on the recording.Some of the involved challenges include selection of electrode material. This has to ensure minimal added thermal noise within the frequency band of LFPs when in contact with cerebrospinal fluid while at the same time being chemically inert and medically harmless. Preliminary results have shown Niobium (Nb) as a promising material suitable for such a recording as it is known to be biologically inert and is commonly used e.g. in dental implants. Its polarizability leads to generally smaller noise power densities in LFP frequency bands than commonly used platinum and tungsten making it a suitable candidate material for neural recordings. This is to be investigated by direct noise measurements in electrolytes and subsequently verified by direct in-vivo measurements.Another challenge lies in the development of acquisition electronics which is greatly constrained by limits imposed on their power consumption governed by safety limits of heat dissipation in neural tissue on the order of 80 mW/cm^2. One of the techniques allowing reduction of used energy is clock-less, also known as continuous-time (CT), acquisition of signals. This approach leads to activity-dependent circuits that only use energy when activity is detected at their input. One of the aims of this project is to investigate the suitability and possible advantages of such circuits for acquisition of LFPs. As properties of such acquisition and sampling processes remain to a large extent unknown, mathematical simulations are used for their investigation. This is complemented by design of integrated circuits and practical measurements verifying hypotheses arising from theoretical models. This has already led to formulation of a minimal requirement to be satisfied in order to prevent aliasing similar to the Nyquist theorem which applies to ordinary sampling. In addition a novel method allowing reduction of flicker noise in CT sampling has been proposed and theoretically validated. An integrated circuit verifying this theory is to be designed, manufactured and tested.This work is part of the ENGINI project with an aim to design the next generation of neural implants that achieve superior chronicity by being completely wireless, minimal in size, targeting LFP recordings and allowing formation of distributed implant networks.The research aligns with the following EPSRC Research Areas: Assistive technology, rehabilitation and musculoskeletal biomechanics; Microelectronic device technology