University:
Boston University
Course:
CAS BI 525 | Biology of Neurodegenerative Diseases
Academic year:

2020
Views:

213

Pages:

17
Author:

Zavier Shaffer

Units, Terms and Standards ш the Reporting of EMG Research : Authors' Note This report is the final report of an ISEK Ad Hoc Committee that was formed in 1977 to deal with the problems arising from inconsistent and erroneous terms and units in the reporting of EMG research. The Committee has addressed this problem at all levels, from the electrophysio- logical terminology right through to the more common processing techniques. Also, the Com- mittee makes some recommendations regarding technical standards that should be aimed for by and from many personal discussions held over the past few years. The document presents not only the fundamental theoretical and physiological relationships but also the total practical ex• perience of the Committee and those consulted. Only a few references have been listed and are intended to represent or amplify certain issues because a complete bibliography of electromyo- graphy would occupy several hundred pages. The Problem It is axiomatic that all researchers in any scientific area should be able to communicate with clarity the results of their work. The area of electromyography is one where inconsistent and erroneous terms and units are the rule rather than the exception. Even veteran researchers are continuing to contribute to the confusion. The research in many major papers cannot be repli- cated because of lack of detail on the protocol, recording equipment or processing technique. For example, the term integrated BMG(IBMG) has been used to describe at least 4 different processing techniques, and the units employed can be mV, mV/sec., mV sec, or just arbitrary units! No wonder there are conflicts and misunderstandings. Form of Report The report concentrates on four major aspects of the myoelectric signal and its subsequent recording and processing. Figure 1 shows the breakdown of the report: Part 1-The Neuromus' cular Domain, Part II-The Recording System, Part III-Temporal Processing and Part IV-Fre- quency Processing. In addition, the Committee presents a Part V-General Experimental Details, which outlines important kinesiological and experimental information that should be reported. References made to the literature are by no means complete, rather they are cited as examples of erroneous or anomalous reporting. Finally, a checklist is given to summarize the recommen- dations of the Committee. In Figure 8(a) is shown a piecewise linear approximation, known as the asymptotic Bode *diagram." This asymptotic diagram is entirely defined by the slopes of the lines, the frequencies where the slope changes, the breakpoints, and a scale factor. The variables discussed so far are commonly used in engineering sciences. Now we will dis- cuss several parameters that have parallels in statistics. In order to do so, the concepts of spec- tral movements will first be defined. The spectral moment of order n is given by: mn " gn w(f)df where f is frequency and W is the power or energy spectrum. The mean frequency, f, is the ratio between the spectral moments or orders one and zero (similar to the mean value in statistics). Thus i = m1 = mo Hz The statistical bandwidth is the square-root of the difference between the ratio of the moment at on er stars i, oto her zero a ththe sou re us the mean, the intensity the stao cro deni equals the square-root of the ratio between the moments of orders two and zero, and the inten- sity of turning points equals the ratio between the fourth and second order moments. The frequency at the maximum of the power spectrum may be called the mode (most prob- able frequency), and the median frequency fm is the frequency which divides the spectrum into two parts of equal power (energy), and is defined by: fm All these parameters, and several not mentioned here, have been used in reports on EMG. research results. The parameters have different properties as they emphasize different aspects of the spectrum. It would be premature to recommend the use of any specitic subset, if such a recommendation should ever be made. It is, however, important to know of the various possi- bilities, and to be familiar with the properties of the parameters. Spectral Estimation Integrals over an infinite time period appear in the mathematical definition of the power spectrum of a noise signal. For obvious reasons then, spectra can never be evaluated exactly; rather any spectrum obtained is an estimate of the mathematical concept. The use of finite stretches of raw EMG data introduces estimation errors. These errors are often summarized as the bias of the estimate, the systematic error, and the variance of the estimate, the statistical uncertainty. Methods employed to reduce these errors include windowing of the raw data, averaging of successive spectral estimates and smoothing of spectral estimates. In reporting results based on spectral analysis, it is important to state the estimation procedure and, if possible, give figures on the bias and the variance of the final estimate. It is important also to report questions pertinent to other sources of errors. In particular, one should give some indica- tion on the noise level of the experimental setup, and state the sampling rate if digital methods are used. The possibility of excessive line interference is easily checked by visual inspection of the specta obtained.