Objective: To demonstrate the errors in surface electromyogram (EMG) interpretation that can be made when the EMG signal is not normalized.
Design: A case study as a proof of principle.
Main outcome measures: The EMG amplitude between the upper and lower portions of the rectus abdominis in one subject during a trunk curl when the EMG signal was normalized (expressed as a percentage of a maximum voluntary contraction) and the amplitude when the signal was expressed in raw, unnormalized arbitrary units or raw millivolts directly read from the instrumentation.
Results: Interpretation of the unnormalized EMG signal suggests that a large difference in neural activation of the upper and lower sections of the rectus abdominis is occurring. In this condition the average activity in the lower rectus is 60.9% of that in the upper portion. This interpretation is incorrect. When the EMG signal is normalized, the differences between the upper and lower segments are reduced. When normalized, the lower segment activity is equal to that of the upper segment.
Conclusions: Because of the inherent EMG signal variability, clinical interpretation of surface EMG signals requires normalization of the signal for physiologic interpretation and for comparison between bilateral muscles and between the same muscle on different days and between different subjects.