针对锋电位的启发式阈值检测算法

Heuristic Threshold Detection Algorithm for Spike

作为植入式脑电信号处理的关键环节,锋电位检测的精确度将直接影响后续脑电信号的解码与分析。提出了一种基于启发式阈值的锋电位自动检测算法。通过对椭圆滤波器参数的优化,降低了原始信号中有用信号衰减程度,实现了较低幅值锋电位的有效保留。并且启发式阈值的设定大大降低了采集环境引入的混杂噪声干扰,实现了具有鲁棒性的锋电位自动阈值检测。基于英国莱斯特大学神经工程实验室提供的细胞外模拟记录数据的实验验证表明,在多种信噪比下提出的算法的平均检测精度可达65.21%。此外,基于猕猴肢体伸展抓握运动范式下采集的植入式脑电数据的实验的结果表明,即使在不确定背景噪声的真实环境中,该算法仍可有效地用于锋电位信号的检测。

Spike detection is the basis of signal processing with implantable brain machine interface system and its detection accuracy affects the accuracy of signal decoding and analysis. An automatic detection algorithm for spikes is proposed.Firstly, elliptic filter parameters are optimized to reduce the attenuation degree of useful signals, so as to retain spikes with low amplitude in the original signals. Then the heuristic threshold formula is used to effectively reduce the mixed noise interference introduced in the complex acquisition environment and realize automatic spike threshold detection.The algorithm is tested on synthetic extracellular records developed at the Neuro-Engineering Lab at the University of Leicester, UK, and shows an average detection accuracy of 65.21% for a variety of SNRS. In addition, spike thresholds are detected in implantable electroencephalogram data collected from the rhesus monkeys’ extended-limb grasping movement paradigm. Experimental results show that the proposed spike detection algorithm can be used to extract spike signals from real data of uncertain background noise.