Coherent self-averaging for the pre-processing of surface electromyography signals in the detection of nociceptive withdrawal reflexes

This work presents an alternative method for the pre-processing of surface electromyography signals of nociceptive withdrawal reflexes, with the goal of improving the reflexes detection in the records. The withdrawal reflex appears when a painful stimulus activates the nociceptors, the generated potential travels to the spinal cord and then the withdrawal of member exposed to painful stimulus is happened. The method used in this study to pre-process these reflexes is called a Coherent Self-Averaging method which attenuates the fluctuations of the random values of high frequency. The smoothing produced by the method is controlled by setting of two parameters: m and k. To demonstrate the performance in the detection of the reflex, 90 reflex signals of 15 healthy subjects were used. Its detection performance was compared with the one of the method called Teager-Kaiser energy operator. The reflex detection performance of algorithms was analyzed using the receiver operating characteristic curve. This showed that the Coherent Self-Averaging method, had a higher sensitivity and specificity than the Teager-Kaiser energy operator. Smoothing of signal, artifact attenuation (by electric stimulation) and reflex enhancement also was observed.