Assessment of central sensitization in chronic pain patients: Novel cross correlation analysis allows withdrawal reflex detection with improved specificity

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

Washington D.C.

Conferencia; Neuroscience 2011; 2011

Recent studies indicate that the human withdrawal reflex might be useful for assessing central sensitization in chronic pain patients. Enlarged reflex receptive field (RRF) has been detected indicating expansion of spinal neuronal receptive fields. The RRF is detected by randomised, distributed electrical stimulation, typically on the sole of the foot. Multiple muscles are activated depending on stimulation site to generate flexion/extension/inversion/eversion as appropriate for the optimal withdrawal. Existing methods for detection of withdrawal reflexes is based on surface electromyography (sEMG) and ignores the possible presence of cross talk from adjacent muscles. In applications assessing RRF this may entail reflex detection from specific muscles with a reduced validity. This study estimated conduction velocities (CV) for tibialis anterior (TA) and soleus (SOL) of both genuine reflexes and cross talk measured during antagonistic reflex responses. Reflexes were elicited by electrical stimulation at either the heel region or on the central pads. Conduction velocity estimations were exploited to develop and assess a novel method for reflex detection robust to cross talk. Cross correlations of two single differential (SD) sEMG signals recorded along the muscle fibres were performed and a set of features including average CV and maximal cross correlation were extracted from the resulting correlogram. Reflex detection based on evaluation of the extracted features where compared to reflex detection based on evaluation of interval peak z-scores proposed by Rhudy et al. (2007). Estimated CV for cross talk were significant higher than CV estimated during genuine reflexes. Reflex detection based on evaluation of interval peak z-scores showed excellent sensitivity (1.00) but poor specificity due to cross talk, particularly when detection was performed on SD sEMG (0.19-0.31). Reflex detection involving evaluation of cross correlations facilitated specificities as high as 0.98 and 0.92 for TA and SOL respectively concurrently with sensitivities of 0.95 and 0.91. The developed method seems not confined to reflex detection but may likely be implemented to ensure a more specific and reliable detection of genuine muscle activation in general. More reliable assessment of withdrawal reflexes is crucial for accurate detection of RRF and thereby quantification of central sensitization in patients with high validity.Rhudy, Pain 128:244-253, 2007