Bruce E N, Goldman M D
Brain Res. 1983 Jun 20;269(2):259-65. doi: 10.1016/0006-8993(83)90135-x.
Electromyograms (EMGs) from respiratory muscles were obtained from human subjects during voluntarily controlled breathing. In 10 studies on 6 subjects EMGs were recorded from the right and left lower ventrolateral surface of the rib cage while the subject emphasized the use of his diaphragm for breathing. Simultaneous samples of the 2 EMG signals of 256-ms duration were obtained once per inspiration (in either the first of second half of inspiration) from each of 30 consecutive breaths using a laboratory minicomputer. Individual power spectra for the 2 EMG signals, and the squared coherence spectrum between them, were calculated. From the coherence spectrum common high-frequency oscillations could be identified within two frequency ranges: 70-100 Hz and 20-50 Hz. Peaks at similar frequencies were identified in both early-inspiratory and late-inspiratory EMG signals. Similar experiments were done on 3 subjects from whom an esophageal diaphragm EMG was obtained as well. The coherence spectrum between the esophageal EMG and the right rib cage surface EMG again demonstrated the presence of common high-frequency oscillations in the same frequency ranges as above. Evidence of these oscillations was not usually apparent in the power spectra. When ECG signal components were allowed to contaminate some of the EMG data samples, the coherence spectrum could be altered considerably, leading to inappropriate conclusions regarding the presence or absence of high-frequency oscillations. It is hypothesized that these high-frequency oscillations are similar to those described in nerve recordings from previous studies on anesthetized and decerebrate cats and dogs and a mechanism for their expression in EMG signals is proposed. The coherence spectrum is a very sensitive method for detecting related signal components in two signals. That high-frequency oscillations can be detected using the less sensitive techniques of autocorrelation and power spectral analysis in some animal preparations may reflect an enhancement of these oscillations in those preparations.
在自主控制呼吸过程中，从人体受试者获取呼吸肌的肌电图（EMG）。在对6名受试者进行的10项研究中，当受试者强调使用膈肌呼吸时，从胸廓左右下外侧表面记录EMG。使用实验室小型计算机，在每次吸气时（吸气的前半段或后半段），从连续30次呼吸中的每一次获取持续时间为256毫秒的两个EMG信号的同步样本。计算两个EMG信号的个体功率谱以及它们之间的平方相干谱。从相干谱中可以在两个频率范围内识别出常见的高频振荡：70 - 100赫兹和20 - 50赫兹。在吸气早期和晚期的EMG信号中都识别出了相似频率的峰值。对另外3名受试者也进行了类似实验，他们还获取了食管膈肌的EMG。食管EMG与右胸廓表面EMG之间的相干谱再次证明在上述相同频率范围内存在常见的高频振荡。这些振荡的证据在功率谱中通常并不明显。当心电图信号成分混入一些EMG数据样本时，相干谱可能会发生很大变化，从而导致关于高频振荡是否存在的不恰当结论。据推测，这些高频振荡与先前对麻醉和去大脑的猫和狗的神经记录中描述的振荡相似，并提出了它们在EMG信号中表达的机制。相干谱是检测两个信号中相关信号成分的非常灵敏的方法。在一些动物实验中，使用不太灵敏的自相关和功率谱分析技术能够检测到高频振荡，这可能反映出这些准备实验中这些振荡有所增强。
