Ferdjallah M, Barr R E
Electrical Engineering Department, University of Texas, Austin 78712.
Biomed Sci Instrum. 1993;29:103-10.
EEG computer analysis is still not widely used in the clinic, and the need for advanced signal processing techniques is still warranted. The fast Fourier transform (FFT) is the method most frequently used for power spectrum estimation of the EEG. In an effort to reduce memory space and processing time required by the fast Fourier transform (FFT), a new method, the enhanced period-peak detection (EPPD), is investigated. The method is based on a combination of the Fourier transform (FT) and period-peak detection. The signal is considered as a train of truncated sinusoidal functions. Each truncated sinusoidal function is limited by two successive local extrema (a peak and a valley). The Fourier transform of the truncated sinusoidal function is the well known Sinc function. The summation of these Sinc functions yields an approximate frequency spectrum of the signal. The speed and performance of the FFT rely upon the number of data collected and the sampling frequency. On the other hand, the enhanced period-peak detection (EPPD) method does not require that the entire EEG data be stored. Only the extrema of the signal and time between the peaks are needed. Furthermore, the frequency resolution of the EPPD is independent of the number of data available and of the sampling frequency.
脑电图计算机分析在临床上仍未广泛应用，因此仍需要先进的信号处理技术。快速傅里叶变换（FFT）是脑电图功率谱估计中最常用的方法。为了减少快速傅里叶变换（FFT）所需的内存空间和处理时间，研究了一种新方法，即增强周期峰值检测（EPPD）。该方法基于傅里叶变换（FT）和周期峰值检测的结合。信号被视为一系列截断的正弦函数。每个截断的正弦函数由两个连续的局部极值（一个峰值和一个谷值）限制。截断正弦函数的傅里叶变换是著名的辛格函数。这些辛格函数的总和产生信号的近似频谱。FFT的速度和性能取决于收集的数据数量和采样频率。另一方面，增强周期峰值检测（EPPD）方法不需要存储整个脑电图数据。只需要信号的极值和峰值之间的时间。此外，EPPD的频率分辨率与可用数据的数量和采样频率无关。
