Nilsson J, Panizza M, Hallett M
Laboratory of Clinical Neurophysiology, Fondazione Clinica del Lavoro, Castel Goffredo (MN), Italy.
Electroencephalogr Clin Neurophysiol. 1993 Oct;89(5):349-58. doi: 10.1016/0168-5597(93)90075-z.
The practice of clinical neurophysiology requires fast, reliable and accurate assessment of a variety of biologic signals. Appropriate filters and rates of analog-to-digital sampling must be used to avoid distorting the signal. Using principles of sampling theory and examples, we describe the frequency content of signals encountered in clinical neurophysiology laboratories, offer guidelines for band-limiting frequencies, and give rules for determining minimal sampling frequencies. Errors introduced by undersampling (aliasing) are illustrated. When sophisticated computational techniques, such as discrete Fourier transform, are used to reconstruct the original wave form, a sampling frequency just above the double of the highest frequency content of the signal is adequate. Sampling neurophysiologic wave forms for direct display, however, requires a sampling frequency at least 4 times as high as that of the upper frequency filter.
临床神经生理学实践需要对各种生物信号进行快速、可靠且准确的评估。必须使用合适的滤波器和模数采样率以避免信号失真。利用采样理论原理及示例，我们描述了临床神经生理学实验室中所遇到信号的频率成分，提供了带宽限制频率的指导原则，并给出了确定最小采样频率的规则。阐述了欠采样（混叠）所引入的误差。当使用诸如离散傅里叶变换等复杂计算技术来重建原始波形时，略高于信号最高频率成分两倍的采样频率就足够了。然而，对神经生理波形进行直接显示采样时，所需的采样频率至少是高频滤波器频率的4倍。
