Recommendations for the use of double differential EMG to record and detect the nociceptive withdrawal reflex in humans

M.B. JENSEN; J.A. BIURRUN MANRESA; O.K. ANDERSEN

San Diego

Conferencia; 6th International IEEE EMBS Neural Engineering Conference (NER); 2013

THE nociceptive withdrawal reflex (NWR) constitutes a valuable surrogate measure to non-invasively probe the excitability of central nociceptive pathways. It is generally elicited by electrical stimulation over the sural nerve or under the sole of the foot and measured over muscles in the lower extremities using electromyography (EMG). To facilitate comparison of results from various studies and to promote clinical usefulness, extensive work has been carried out to achieve reliable standardized scoring criteria for detection of the NWR from EMG. Among other criteria, the interval peak z-score has been reported as being highly accurate and reliable [1]. We have recently documented that the evaluation of standard single differential (SD) EMG does not allow reliable muscle-specific reflex detection due to the volume conduction of interfering myoelectric signals from nearby muscles [2]. The negative implications of this phenomenon, denoted crosstalk, may be significantly reduced by the use of double differential (DD) EMG recordings. However, it has not yet been reported how the use of a different recording technique influences the cut-point of the standardized scoring criteria. Methods: Reflexes were elicited by electrical stimulation at the sole of the foot and measured using both SD and DD surface (sEMG) and intramuscular (iEMG) EMG from the tibialis anterior and soleus muscles in 15 healthy subjects. The interval peak z-score were calculated for 489 individual recordings. Two sets of ROC-curves were produced for both SD and DD sEMG respectively using two different gold standards; 1) assessing sEMG only and 2) assessing both sEMG and iEMG (enabling distinction between genuine reflexes and crosstalk).Results: The ROC-curves produced considering gold standard 1 indicated that evaluation of interval peak z-scores calculated for both SD and DD sEMG allows reflex detection with high accuracy with areas under the curve (AUC) of 0.99 and 0.98 respectively. The corresponding sensitivity and specificity plots indicated optimal cut-points at z=12.6 and z=14.3 for SD and DD sEMG, respectively. However, the ROC-curves produced considering gold standard 2 revealed a significant difference in accuracy between the evaluation of SD (AUC=0.95) and DD (AUC=0.99) sEMG (p