A method for quantifying reflex responses from intra-muscular and surface electromyogram.

Measuring human reflex responses from electromyogram (EMG) traces in an accurate, repeatable and reliable way with a high degree of specificity has traditionally been a difficult task. This paper describes a new method that can be used to quantify reflex responses from both surface and intra-muscular EMG. This technique extends the classical cumulative sum (CUSUM) calculations by defining precise points for the calculation of latencies, durations and strengths to facilitate automatic reflex detection and permit the strength of a reflex to be defined in absolute units. The effect of varying the pre-stimulus time, the number of trials averaged and the amount of filtering used on the identification and classification of reflex parameters are also investigated. Furthermore, the effect of noise on these values, and how to remove it, is discussed. The new method, which is an expansion of the CUSUM analysis, is compared and contrasted with the more common threshold-crossing method in two different muscles: masseter and first dorsal interosseous (FDI), in experiments utilizing both mechanical and electrical stimulation. There are a number of advantages to using the new method; not only does the modified CUSUM method detect reflexes earlier than threshold-crossing methods but also the strength and duration are less susceptible to averaging and filtering parameters while giving a better indication of the reflex size. The data suggests that a pre-stimulus analysis period of at least 100 ms be used to correctly identify the variability inherent in EMG traces. It is also concluded that for subtle reflexes, 50 stimuli should be the minimum number used when spike trigger averaging is employed as lower numbers are associated with much greater pre-stimulus variability. Zero-phase filtering the rectified averaged EMG traces is recommended as this makes it easier to identify significant changes in the electrical activity of the muscle in question. In addition, noise estimation and removal from averaged rectified EMG recordings yields results that are a more accurate representation of the synaptic activity of the motor units in question.