诱发电位快速提取方法及进展

Methods and Algorithms for Fast Extraction of Evoked Potentials

每次刺激所产生的诱发电位的峰值和潜伏期以至波形是时变信号。随着信号处理技术的发展,诱发电位的提取方法由传统的平均叠加方法,向逐步减少累加次数,最终实现单次提取诱发响应波形,实现动态提取诱发电位的方向发展,因此对诱发电位的少次以至单次提取成为生物医学信号处理领域备受关注的一个研究课题。对近年来快速提取诱发电位的信号处理方法进行了简要的回顾,并分别从自适应滤波、小波变换、神经网络及独立成分分析等4个方面对算法进行了介绍。

Evoked potentials（EPs） are nonstationary and have characteristics that vary throughout stimulus trails. With the development of signal processing, new techniques focus on using fewer trails to extract the evoked po-tentials and to achieve single-trial extraction of evoked potentials finally. Therefore, fewer trials or single-trial estimation of evoked potentials has become the hotspot in the field of biomedical signal processing. In this article, signal processing methods of fast extraction of evoked potentials developed in recent years are reviewed and the approaches based on adaptive falter, wavelet transform, neural network and independent component analysis are discussed.