Classification of large array surface myoelectric potentials from subjects with and without low back pain

Abstract

An algorithm was developed and tested for differentiating between the spatial distribution of large arrays of surface electromyographic (LASE) data from subjects with and without low back pain (LBP). The surface EMG data from 62 channels were collected from the low back of 161 healthy and 44 acute (less than 6-weeks) LBP subjects in three minimum stress postural positions including standing, 20 degrees of trunk flexion (at hip joint) and standing with arms extended forward holding a 1.36 kg (3 lb) weight in each hand. These data were statistically analyzed and the spatial distribution of the root mean square (RMS) values was used in a multivariate quadratic discriminant model to reclassify the healthy and acute LBP subjects. The most predictive results were obtained from the ‘flexion’ group of experiments and correctly reclassified 95.5% (42/44) of the acute LBP subjects and 99.4% (160/161) of the healthy subjects. The success rate of this reclassification based on surface distribution of myoelectric potentials was found to be better than the reported patient classifications based on a smaller set of electrode pairs using fewer subjects [Peach JP, McGill SM, Classification of low back pain with use of spectral electromyogram parameters. Spine 23(10):1998;1117–23; Roy SH, De Luca CJ, Emley M, Oddsson LI, Buijs RJ, Levins JA, Newcombe DS, Jabre JF. Classification of back muscle impairment based on the surface electromyographic signal. J Rehabil Res Dev 34(4):1997;405–14 [review]]. The results indicated the potential of the model for clinical patient classification.