基于数学形态学与核主成分分析的峰电位检测与分类方法

摘要

Objective We introduce a new unsupervised method for detecting and sorting spikes from extracellular recordings. Methods First, multiple mathematical morphology operation is used in signal de-noising before spike detection with a fixed threshold. Then,wavelet transform and kernel principal components analysis ( KPCA ) are performed to the detected spike waveforms to extract discriminative features. Finally, the minimum-distance clustering is proceeded to sort spikes. Results The simulation experimental results indicate that the spike detectable rate is 94% . The classification accuracy in general is over 91% and that with many superposed signals is over 88% . Conclusions The results show that the method performs quite well even with the noisy simulated spike data.%目的 为抑制高强度背景噪声及信号叠加的干扰,提高峰电位的检出率和分类的正确性,本文提出一种新的无监督方法.方法 首先,应用数学形态学的复合操作对信号进行降噪,采用定阈值提取峰电位.然后,小波变换和核主成分分析法(kernel principal components analysis,KPCA)相结合,对已提取的峰电位波形进行特征提取.最后,用改进的最小距离法实现峰电位分类.结果 仿真实验结果表明,此方法对于不同噪声强度的信号,峰电位检出率达94%,总分类正确率91%以上,其中大量叠加信号的分类正确率88%以上.结论 本方法能在有效抑制噪声的基础上,准确提取峰电位并有效分类.