622 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 52, NO. 4, APRIL 2005 A Technique to Track Individual Motor Unit Action Potentials in Surface EMG by Monitoring Their Conduction Velocities and Amplitudes Rebecca B. J. Beck*, Caroline J. Houtman, Mark J. O’Malley, Senior Member, IEEE, Madeleine M. Lowery, Member, IEEE, and Dick F. Stegeman, Member, IEEE Abstract—The speed of propagation of an action potential along a muscle fiber, its conduction velocity (CV), can be used as an indication of the physiological or pathological state of the muscle fiber membrane. The motor unit action potential (MUAP), the waveform resulting from the spatial and temporal summation of the individual muscle fiber action potentials of that motor unit (MU), propagates with a speed referred to as the motor unit con- duction velocity (MUCV). This paper introduces a new algorithm, the MU tracking algorithm, which estimates MUCVs and MUAP amplitudes for individual MUs in a localized MU population using SEMG signals. By tracking these values across time, the electrical activity of the localized MU pool can be monitored. An assessment of the performance of the algorithm has been achieved using sim- ulated SEMG signals. It is concluded that this analysis technique enhances the suitability of SEMG for clinical applications and points toward a future of noninvasive diagnosis and assessment of neuromuscular disorders. Index Terms—Motor unit conduction velocity, peak velocity, surface electromyography. I. INTRODUCTION T HE conduction velocity (CV) of an action potential along a muscle fiber is intimately linked to properties of the muscle fiber membrane. CV estimates may, therefore, be used as an indication of the physiological or pathological state of the muscle. A common application of CV is the assessment of muscle fatigue [1], [2]. The standard approach to CV estima- tion from the surface electromyogram (SEMG) is to determine the time taken for the signal (typically 0.5 to 2 s in duration), to travel between two spatially displaced recording points. A single CV value is determined from the quotient of the distance travelled and the time taken. Such an “average” or “global” es- timate is referred to here as the global-CV estimate. The SEMG Manuscript received May 16, 2003; revised September 12, 2004. Asterisk indicates corresponding author. *R. B. J. Beck is with the Department of Electronic and Electrical Engineering, University College Dublin, Dublin 4, Ireland (e-mail: re- becca.beck@ucd.ie). C. J. Houtman and D. F. Stegeman are with the Department of Clinical Neuro- physiology, Institute of Neurology, University Medical Centre Nijmegen, 6500 HB Nijmegen, The Netherlands. They are also with the Institute of Funda- mental and Clinical Human Movement Sciences, Amsterdam and Nijmegen, The Netherlands. M. J. O’Malley is with the Department of Electronic and Electrical Engi- neering, University College Dublin, Ireland (e-mail: mark.omalley@.ucd.ie). M. M. Lowery is with the Sensory Motor Performance Program, Rehabili- tation Institute of Chicago, IL 60613 USA. She is also with the Department of Physical Medicine and Rehabilitation, Northwestern University, IL 60611 USA. Digital Object Identifier 10.1109/TBME.2005.844027 generated from a voluntarily contracting muscle comprises the combined electrical activity of a number of different MUs with different fiber types, varying sizes and relative locations, dif- ferent fatigue and recruitment patterns and different motor unit conduction velocities (MUCVs) and firing statistics. Therefore, a single, global-CV estimate, may be of limited value. By con- trast, the distributions of muscle fiber CVs, obtained from the analysis of needle electromyograms (NEMGs), have proved to be of important diagnostic value [3]–[5]. Researchers in the field of SEMG have, therefore, been motivated to develop alternative, more informative means of MUCV estimation from SEMG. A number of approaches have been taken in response to this challenge. Some have chosen to extract individual MUAPs from the SEMG prior to estimating the corresponding MUCVs [6]–[8]. Others have approached the problem by way of decon- volution analysis, where the distal SEMG signal is viewed as a convolution of the proximal SEMG signal with an impulse response filter, which represents the delays due to propagation [9], [10]. Though such an approach has a theoretical grounding, it is not yet appropriate for the analysis of real signals [9], [11]. CV values have also been determined for normalized average peaks, generated from a minimum of 500 individual peaks in the SEMG signal [12]. However, such estimates would provide little or no information about the actual CVs of the underlying muscle fibers. An alternative and increasingly common approach to CV estimation from SEMG—the peak velocity approach—estimates the velocities of individual peaks in the SEMG signal [11], [13]–[15]. However, all peaks in the SEMG signal may not correspond to individual MUAPs, due to MUAP interference. Therefore, all peak velocity values in the distributions do not necessarily represent the underlying MUCVs. Such peak velocity distributions may also be biased toward the velocities of MUs with higher firing rates. An electrode combination with a large pick-up area may en- able the activity of a large number of MUs to be recorded. The resultant signals will tend to be complicated, i.e., probable MUAP interference, and hence difficult to analyze, resulting in less reliable MUCV estimates. A more selective electrode com- bination could result in clearer signals and hence more accurate MUCV estimates, though the number of estimates or the size of the resultant distribution may be reduced. Any approach to MUCV distribution estimation from SEMG, therefore, inher- ently contains a tradeoff between the number of MUs examined and the level of accuracy of the results [11]. Certain common issues must be dealt with when estimating MUCV distributions 0018-9294/$20.00 © 2005 IEEE