Estimating motor unit discharge patterns from high-density surface electromyogram Aleš Holobar a,b , Dario Farina c , Marco Gazzoni b , Roberto Merletti b, * , Damjan Zazula a a Faculty of Electrical Engineering and Computer Science, University of Maribor, Maribor, Slovenia b Laboratorio di Ingegneria del Sistema Neuromuscolare, Politecnico di Torino, Dipartimento di Elettronica, Corso Duca degli Abruzzi, 24, 10129 Torino, Italy c Center for Sensory–Motor Interaction (SMI), Department of Health Science and Technology, Aalborg University, Aalborg, Denmark See Editorial, pages 449–450 article info Article history: Accepted 28 October 2008 Available online 8 February 2009 Keywords: Motor unit Discharge pattern High-density EMG Surface EMG Decomposition abstract Objective: We systematically tested the capability of the Convolution Kernel Compensation (CKC) method to identify motor unit (MU) discharge patterns from the simulated and experimental surface electromyogram (sEMG) during low-force contractions. Methods: sEMG was detected with a grid of 13  5 electrodes. In simulated signals with 20 dB signal-to- noise ratio, 11 ± 3 out of 63 concurrently active MUs were identified with sensitivity >95% in the estimation of their discharge times. In experimental signals recorded at 0–10% of the maximal force, the discharge pat- terns of (range) 11–19 MUs (abductor pollicis; n = 8 subjects), 9–17 MUs (biceps brachii; n = 2), 7–11 MUs (upper trapezius; n = 2), and 6–10 MUs (vastus lateralis; n = 2) were identified. In the abductor digiti minimi muscle of one subject, the decomposition results from concurrently recorded sEMG and intramuscular EMG (iEMG) were compared; the two approaches agreed on 98 ± 1% of MU discharges. Conclusion: It is possible to identify the discharge patterns of several MUs during low-force contractions from high-density sEMG. Significance: sEMG can be used for the analysis of individual MUs when the application of needles is not desirable or in combination with iEMG to increase the number of sampled MUs. Ó 2009 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved. 1. Introduction Recording of the electrical activities of motor units (MUs) pro- vides an insight into the activation properties of the motoneurons that are located in the spinal cord (De Luca et al., 1982). However, electric signals detected from muscles comprise the contributions of all the MUs which are active within the detection volume of the recording system. Analysis of MU discharges thus requires automated signal decomposition (De Luca and Adam, 1999; McGill and Dorfman, 1985; Stashuk, 2001) which attempts to resolve a composite EMG signal into its constituent MU action potential (MUAP) trains. Identification of individual MU electrical activities in vivo is classically performed by intramuscular EMG (iEMG), which has high spatial selectivity and thus comprises the contributions of a relatively small number of active MUs whose fibers are close to the recording point (De Luca and Adam, 1999; McGill and Dorfman, 1985; Stashuk, 2001; Nawab et al., 2008). Because this method is invasive, it has limitations in cases when needle insertion is not possible or not desirable, such as in clinical examinations of chil- dren, professional athletes, patients with transplanted limbs (Fari- na et al., 2008a) or haemophilia, and, in general, in dynamic conditions, during work, sport or space activities. Moreover, the identified intramuscular action potentials are not representative of all the fibers in the MU (Stålberg, 1980) and it is not possible to detect the same MUs in repeated measurements. With indwell- ing EMG recordings, it is also difficult to extract parameters related to the membrane properties of the muscle fibers, such as action po- tential propagation velocity (Farina et al., 2001; Merletti, 1994), and anatomical characteristics of the MUs, such as fiber length, fi- ber orientation, and location of the innervation zone (Merletti, 1994). The information on membrane and anatomical fiber proper- ties is relevant in several applications, e.g., for studying muscle fa- tigue or the effect of location for the injection of substances such as botulin toxin. 1388-2457/$36.00 Ó 2009 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.clinph.2008.10.160 * Corresponding author. Tel.: +39 011 4330476; fax: +39 011 4330404. E-mail address: roberto.merletti@polito.it (R. Merletti). Clinical Neurophysiology 120 (2009) 551–562 Contents lists available at ScienceDirect Clinical Neurophysiology journal homepage: www.elsevier.com/locate/clinph