A Pilot Clinical Trial for Speech Neuroprosthesis

神经损伤导致的沟通能力丧失是毁灭性的。这项拟议的试点试验旨在评估语音假肢技术在一小群瘫痪患者中的可行性。我们将利用对人类语言运动皮层功能特性长达十年的科学研究，以及最近在正常说话的患者中植入电极（用于癫痫手术计划）的语音解码演示。我们建议早期可行性研究的长期直接神经接口恢复通信。为了实现这一目标，我们提出了三个关键的创新：1)长期植入高密度128通道皮质电图（ECoG）阵列，从整个外侧运动皮层（包括言语运动皮层）取样；2)应用最先进的神经网络解码方法；3)直接比较文本和语音合成解码方法。皮质电图是一种使用非穿透电极记录大脑表面神经活动（局部场电位）的方法。我们假设ECoG在临床应用方面可能比目前的替代方案（如微电极或头皮脑电图）具有明显的优势，因为我们既可以实现高密度采样，又可以覆盖整个言语运动皮层。我们之前已经证明，这种规模和覆盖范围是必要的，足以解码完整个体的语音。ECoG在癫痫和脑机接口等人类医疗应用中长期植入的安全性和可靠性方面有越来越多的证据。我们已经消除了FDA IDE和当地IRB批准用于研究用途的监管障碍。该研究的主要目标是通过神经信号解码的文本和合成语音实现通信。第二个目标是证明慢性植入情况下基于心电图记录的鲁棒性和稳定性。这些目标将决定语言神经假肢技术在瘫痪患者目标人群中的可行性。

Neurological injury that results in the loss of communication is devastating. This proposed pilot trial is designed to evaluate the feasibility of speech prosthetic technology in a small cohort of paralyzed patients. We will leverage a decade-long scientific investigation of the functional properties of human speech motor cortex, and recent demonstration of speech decoding in normally speaking patients implanted with electrodes (for epilepsy surgery planning). We propose an early feasibility study of a long-term direct neural interface for restoration of communication. We propose three critical innovations to achieve this goal: 1) a chronically implanted high-density 128 channel electrocorticography (ECoG) array which samples from the entire lateral motor cortex (including speech motor cortex), 2) application of state-of-the-art neural network decoding approaches, and 3) direct comparison of text and speech synthesis decoder approaches. Electrocorticography is a method of recording neural activity (local field potentials) from the brain surface using non-penetrating electrodes. We hypothesize that ECoG may have distinct advantages for clinical application over current alternatives (e.g. microelectrode or scalp EEG) given that we can achieve both high-density sampling and cover the entire speech motor cortex. We have previously demonstrated that this scale and coverage is necessary and sufficient to decode speech in intact individuals. ECoG has an increasing well documented safety and reliability profile for long-term implantation in human medical applications such epilepsy and brain computer interfaces. We have already de-risked regulatory hurdles with FDA IDE and local IRB approval for investigational use. The primary goals of the study are to enable communication via text and synthesized speech decoded from neural signals. The secondary goal is to demonstrate the robustness and stability of ECoG-based recordings in the chronic implantation setting. These aims will determine the feasibility of speech neuroprosthetic technology in target population of patients who are paralyzed.