Ashmore Robin C, Endler Bridget M, Smalianchuk Ivan, Degenhart Alan D, Hatsopoulos Nicholas G, Tyler-Kabara Elizabeth C, Batista Aaron P, Wang Wei
University of Pittsburgh, Pittsburgh, PA, USA.
Annu Int Conf IEEE Eng Med Biol Soc. 2012;2012:1740-4. doi: 10.1109/EMBC.2012.6346285.
A brain computer interface (BCI) system was implemented by recording electrocorticographic signals (ECoG) from the motor cortex of a Rhesus macaque. These signals were used to control two-dimensional cursor movements in a standard center-out task, utilizing an optimal linear estimation (OLE) method. We examined the time course over which a monkey could acquire accurate control when operating in a co-adaptive training scheme. Accurate and maintained control was achieved after 4-5 days. We then held the decode parameters constant and observed stable control over the next 28 days. We also investigated the underlying neural strategy employed for control, asking whether neural features that were correlated with a given kinematic output (e.g. velocity in a certain direction) were clustered anatomically, and whether the features were coordinated or conflicting in their contributions to the control signal.
通过记录恒河猴运动皮层的脑电信号（ECoG）实现了一个脑机接口（BCI）系统。这些信号被用于在标准的中心外任务中控制二维光标移动，采用的是最优线性估计（OLE）方法。我们研究了猴子在协同自适应训练方案中操作时获得精确控制的时间进程。4 - 5天后实现了精确且持续的控制。然后我们保持解码参数不变，并在接下来的28天观察到稳定的控制。我们还研究了用于控制的潜在神经策略，询问与给定运动输出（如特定方向的速度）相关的神经特征在解剖学上是否聚类，以及这些特征对控制信号的贡献是协同的还是相互冲突的。
