A Robust Encoding Scheme for Delivering Artificial Sensory Information via Direct Brain Stimulation
A Robust Encoding Scheme for Delivering Artificial Sensory Information via Direct Brain Stimulation
复制标题
DOI：
10.1109/tnsre.2019.2936739
复制
发表时间：
2019-10-01
影响因子：
4.9
通讯作者：
Moritz, Chet T.
中科院分区：
文献类型：
--
作者：
Bjanes, David A.;Moritz, Chet T.
下一代双向皮质神经假体需要通过直接电刺激大脑来创造体感的创新。目前触觉感知和本体感受输入的缺乏可能会对脑控假肢或复活肢体的速度和准确性造成根本限制。这项研究解决了在高维参数空间中识别鲁棒的，高带宽的感官编码方案的独特挑战。之前的研究证明了一维编码方案可以提供低带宽的感觉信息，但没有对参数进行比较，也没有与适合神经假体实时操作的更新速率进行比较。在这里，我们报告了对关键刺激参数（如脉冲幅度、脉冲宽度、频率、脉冲串间隔和脉冲数）的分辨率的首次全面测量。令人惊讶的是，刺激频率的调制在很大程度上是不可检测的。虽然我们最初预计高频内容是将高通量感觉信号传递到大脑的理想候选者，但我们发现只有非常低的频率的调制才能检测到。相反，每个脉冲的每相电荷产生最高分辨率的感觉信号，并且是调制感知强度的关键参数。刺激编码模式设计用于双向大脑接口所需的高带宽信息传输。我们发现的刺激功能，最好的编码感知强度有显着的影响，任何神经接口的设计，寻求通过电刺激直接向大脑传递信息。
Innovations for creating somatosensation via direct electrical stimulation of the brain will be required for the next generation of bi-directional cortical neuroprostheses. The current lack of tactile perception and proprioceptive input likely imposes a fundamental limit on speed and accuracy of brain-controlled prostheses or re-animated limbs. This study addresses the unique challenge of identifying a robust, high bandwidth sensory encoding scheme in a high-dimensional parameter space. Previous studies demonstrated single dimensional encoding schemes delivering low bandwidth sensory information, but no comparison has been performed across parameters, nor with update rates suitable for real-time operation of a neuroprosthesis. Here, we report the first comprehensive measurement of the resolution of key stimulation parameters such as pulse amplitude, pulse width, frequency, train interval and number of pulses. Surprisingly, modulation of stimulation frequency was largely undetectable. While we initially expected high frequency content to be an ideal candidate for passing high throughput sensory signals to the brain, we found only modulation of very low frequencies were detectable. Instead, the charge-per-phase of each pulse yields the highest resolution sensory signal, and is the key parameter modulating perceived intensity. The stimulation encoding patterns were designed for high-bandwidth information transfer that will be required for bi-directional brain interfaces. Our discovery of the stimulation features which best encode perceived intensity have significant implications for design of any neural interface seeking to convey information directly to the brain via electrical stimulation.