中文摘要：

多通道在体记录可以同时记录到多个神经元的胞外放电信号以及对应的局部场电位的活动信号。如何对记录到的这两种电信号进行合适的处理，以确保实验结果的准确性，是运用好多通道在体记录技术的关键之一。本文旨在针对多通道在体记录的原始数据，介绍动作电位及场电位信号的常用数据处理方法。动作电位信号属于高频信号，一般用40 kHz的高速采样频率进行采集和记录。根据记录到的神经元胞外动作电位波形，运用主成分分析技术，再结合四电极记录技术的优势，可对来自记录电极周围不同空间位置的神经元放电信号进行良好的甄别，从而获得较精确的单神经元放电时间序列。而局部场电位信号属低频信号（〈 300 Hz），一般用1 kHz的采样频率进行采集和记录。记录到的场电位原始信号需要进行数字滤波，从而分离出场电位信号中不同频率段的节律性振荡。啮齿类动物海马结构中常见的节律性振荡有动物清醒活动及快速眼动睡眠时的 theta节律（4-12 Hz）；清醒认知活动过程中，伴随着theta节律一起出现的gamma节律（30-80 Hz）；以及清醒静止及慢波睡眠时的ripple高频振荡（100-250 Hz）。针对以上处理获得的数据，常用的后续数据分析方法有：神经元放电间隔分析、神经元放电自相关与互相关分析、以及信号的频谱分析等。

英文摘要：

Multi-channel in vivo recording techniques are used to record ensemble neuronal activity and local field potentials （LFP） simultaneously. One of the key points for the technique is how to process these two sets of recorded neural signals properly so that data accuracy can be assured. We intend to introduce data processing approaches for action potentials and LFP based on the original data collected through multi-channel recording system. Action potential signals are high-frequency signals, hence high sampling rate of 40 kHz is normally chosen for recording. Based on waveforms of extracellularly recorded action potentials, tetrode technology combining principal component analysis can be used to discriminate neuronal spiking signals from differently spatially distributed neurons, in order to obtain accurate single neuron spiking activity. LFPs are low-frequency signals （lower than 300 Hz）, hence the sampling rate of 1 kHz is used for LFPs. Digital filtering is required for LFP analysis to isolate different frequency oscillations including theta oscillation （4-12 Hz）, which is dominant in active exploration and rapid-eye-movement （REM） sleep, gamma oscillation （30-80 Hz）, which is accompanied by theta oscillation during cognitive processing, and high frequency ripple oscillation （100-250 Hz） in awake immobility and slow wave sleep （SWS） state in rodent hippocampus. For the obtained signals, common data post-processing methods include inter-spike interval analysis, spike auto-correlation analysis, spike cross-correlation analysis, power spectral density analysis, and spectrogram analysis.