Kang Xiaoxu, Schieber Marc, Thakor Nitish V
Department of Biomedical Engineering, Johns Hopkins University, Baltimore, MD.
Department of Neurology, Cognitive Behavioral Neurology, University of Rochester, Rochester, NY.
Proc Int Conf Neural Inf Process. 2012 Nov;7663:68-75. doi: 10.1007/978-3-642-34475-6_9.
Advanced dexterous prosthetics technology has been under rapid development as a potential solution to upper limb amputation. An important problem in the neural prosthetics design is to develop control policies for prosthesis movements. This requires an estimation of cognitive states. Previous works mostly used premotor and primary motor neurons to estimate cognitive states. Here we demonstrate that the recorded neural activity from the somatosensory cortex can be used to estimate cognitive states in complex behavioral tasks. We measure the latencies between the predicted cognitive state transitions and true transitions. The maximum prediction latency for grasping a sphere, pulling a mallet, pushing a button and pulling a cylinder are 42±21 ms, 91.6±10.7 ms, 177.1±94.6 ms, 22.5±74.5 ms, respectively. These latency estimates indicate that good timing of the cognitive states with small latencies can be obtained from the somatosensory neural data to plan movements of prosthetic limbs.
先进的灵巧假肢技术作为上肢截肢的一种潜在解决方案一直在快速发展。神经假肢设计中的一个重要问题是制定假肢运动的控制策略。这需要对认知状态进行估计。先前的工作大多使用运动前神经元和初级运动神经元来估计认知状态。在这里，我们证明了从体感皮层记录的神经活动可用于估计复杂行为任务中的认知状态。我们测量预测的认知状态转换与真实转换之间的延迟。抓握球体、拉木槌、按按钮和拉圆柱体的最大预测延迟分别为42±21毫秒、91.6±10.7毫秒、177.1±94.6毫秒、22.5±74.5毫秒。这些延迟估计表明，从体感神经数据中可以获得具有小延迟的认知状态的良好时机，以规划假肢的运动。
