- Email: [email protected]
Validation of a portable EMG device to assess muscle activity during free-living situations
Journal of Electromyography and Kinesiology xxx (2013) xxx–xxx
Contents lists available at SciVerse ScienceDirect
Journal of Electromyography and Kinesiology journal homepage: www.elsevier.com/locate/jelekin
Validation of a portable EMG device to assess muscle activity during free-living situations T.J. Walters, K.A. Kaschinske, S.J. Strath, A.M. Swartz, K.G. Keenan ⇑ University of Wisconsin-Milwaukee, Department of Kinesiology, Milwaukee, WI, USA
a r t i c l e
i n f o
Article history: Received 14 November 2012 Received in revised form 9 May 2013 Accepted 8 June 2013 Available online xxxx Keywords: Activities of daily living Movement Motor activity Muscle strength
a b s t r a c t Portable amplifiers that record electromyograms (EMGs) for longer than four hours are commonly priced over $20,000 USD. This cost, and the technical challenges associated with recording EMGs during freeliving situations, typically restrict EMG use to laboratory settings. A low-cost system (lEMG; OT Bioelecttronica, 100€), using specialized concentric bipolar electrodes, has been developed specifically for free-living situations. The purpose of this study was to validate the lEMG system by comparing EMGs from lEMG with a laboratory-based alternative (Telemyo 900; Noraxon USA, Inc.). Surface EMGs from biceps brachii (BB) and tibialis anterior (TA) of ten subjects were recorded simultaneously with both systems as subjects performed maximal voluntary contractions (MVCs), submaximal contractions at 25%, 50%, and 75% MVC, seven simulated activities of daily living (ADLs), and >60 min of simulated free-living inside the laboratory. In general, EMG parameters (e.g., average full-wave rectified EMG amplitude) derived from both systems were not significantly different for all outcome variables, except there were small differences across systems in baseline noise and absolute EMG amplitudes during MVCs. These results suggest that lEMG is a valid approach to the long-term recording of EMG. Ó 2013 Published by Elsevier Ltd.
1. Introduction Accurate measurement of neuromuscular function in free-living situations is critical to assess the functional use of muscles during activities of daily living (ADLs) and has been used to investigate work-related pain (Ostensvik et al., 2009), advancing age (Theou et al., 2010), and clinical disorder (Jakobi et al., 2008). Although movement during free-living situations is most commonly studied using questionnaires, pedometers, and accelerometers, only electromyography (EMG) can indicate the neuromuscular strategy underlying movement during daily activities. Unfortunately, there are a number of technical challenges associated with using laboratory-based, portable EMG systems during free living. These challenges include spurious artifacts and noise in long-term EMG recordings that arise from movement of the EMG system and cabling when used across different environments (Jakobi et al., 2008; Klein et al., 2010). These technical challenges, as well as the high cost of conventional EMG systems, typically restrict their use to laboratory settings and long-term EMG recordings from human muscles are still relatively scarce (e.g., Jakobi et al., 2008; Kern et al., 2001; Klein et al., 2010; Monster et al., 1978; Mork and
⇑ Corresponding author. Address: The University of Wisconsin-Milwaukee, 2400 E. Hartford Ave., Milwaukee, WI 53211, USA. Tel.: +1 (414) 229 2336; fax: +1 (414) 229 2619. E-mail address: [email protected] (K.G. Keenan).
Westgaard, 2005; Ostensvik et al., 2009). Alternatively, low-cost EMG data logging systems have been developed specifically for use during free living, though the validity of these low-cost EMG systems has not been examined. One system developed specifically for recording EMGs during free living is lEMG (OT Bioelettronica, Torino, Italy). In contrast to typical laboratory-based EMG systems, lEMG is inexpensive (100€), small (57 mm 36 mm 16 mm), and lightweight (25 g). Similar to hip-worn accelerometers commonly used to estimate energy expenditure and track physical activity outside the laboratory (Swartz et al., 2000; Troiano et al., 2008), the low-profile lEMG is easily positioned on the subject and can be deployed from any convenient location to collect EMG data continuously during free-living situations for up to nine hours. Since laboratory-based EMG systems typically cost over $20,000 USD, the relative low-cost and specific design of lEMG (as highlighted below) may make it a practical solution to larger-scale, out-of-laboratory data collections. However, there are substantial differences between lEMG and laboratory-based systems that warrant validation of lEMG. Potential limitations of the lEMG system include: one amplification gain (980 V/V), a limited output range of 3.3 V, and an A/D resolution of 8 bit. Thus, the resolution of lEMG is limited compared with laboratory-based systems and more susceptible to signal saturation which could decrease its usefulness. Moreover, it is unclear how lEMG is influenced by different sources of noise and signal artifact
1050-6411/$ - see front matter Ó 2013 Published by Elsevier Ltd. http://dx.doi.org/10.1016/j.jelekin.2013.06.004
Please cite this article in press as: Walters TJ et al. Validation of a portable EMG device to assess muscle activity during free-living situations. J Electromyogr Kinesiol (2013), http://dx.doi.org/10.1016/j.jelekin.2013.06.004
2
T.J. Walters et al. / Journal of Electromyography and Kinesiology xxx (2013) xxx–xxx
that can be common in free-living situations (De Luca et al., 2010; Klein et al., 2010). In contrast to these possible limitations, lEMG has features that may enhance its practicality for use in free-living situations compared with laboratory-based systems. Specifically, lEMG uses concentric detection bipolar lead electrodes which are isotropic, or invariant to rotations of the electrode with respect to the underlying muscle fiber orientation (Farina and Cescon, 2001). Consequently, the concentric electrodes can be placed without consideration of muscle fiber orientation or changes in muscle fiber orientation that typically occur during anisometric contractions (Farina and Cescon, 2001). Furthermore, concentric electrodes are more selective and thus less susceptible to the problem of crosstalk (Farina and Cescon, 2001). In addition, the concentric electrode developed by OT Bioelettronica has a low-profile design and large adhesive surface that is unlikely to be dislodged after hours of wear, as well as short cable lengths that may minimize artifacts. Nonetheless, although the advantages of surface concentric electrodes have been well described theoretically using numerical models (Farina and Cescon, 2001), no experimental studies have yet used the surface concentric electrodes. The purpose of this study was to validate lEMG by comparing EMGs from lEMG with simultaneous recordings from Noraxon Telemyo 900. The study involved having subjects perform a range of isometric forces, seven simulated ADLs and a short (>1 h) simulated free-living condition. EMGs were collected simultaneously with both EMG systems from biceps brachii (BB) and tibialis anterior (TA) and EMG parameters were compared across systems.
2. Methods 2.1. Subjects Ten healthy subjects (5 males, 24.5 ± 7.4 yrs (mean ± SD); range: 19–44 yrs old) participated in the study. All subjects
reported to be healthy and free of known orthopedic or neuromuscular disorders. All participants read and signed an informed consent document approved by the University Institutional Review Board prior to participation. 2.2. Setup Table 1 highlights specifications for both lEMG and the laboratory-based system (Telemyo 900, Noraxon USA, Inc.) used for comparison in the current study. EMGs were collected simultaneously with both EMG systems with electrodes positioned on the skin overlying biceps brachii (BB) and tibialis anterior (TA). BB and TA were selected as they are relatively long muscles with clearly identified innervation zones and tendon endings (Rainoldi et al., 2004; Saitou et al., 2000); thus, placing electrodes from both systems on either side of the innervation zone was possible (Fig. 1). In addition, both muscles have relatively low pennation angles and are less influenced by crosstalk because of their relatively isolated anatomical positions, thus eliminating advantages of the lEMG system and allowing the Noraxon Telemyo 900 to provide a more realistic criterion for comparison to lEMG.
lEMG: Each subject wore two lEMGs with concentric electrodes. Concentric electrodes consisted of Ag/AgCl with a circular point electrode (1.6 cm diameter) and surrounding external circular ring (outside and inside diameter of 5 cm and 4 cm, respectively) and were placed on the skin overlying BB and TA on the right side of the body. lEMG units were secured and grounded with square adhesive electrodes (10 10 mm; Vermed, Bellow Falls, VT) over the acromio-clavicular joint and over the tibial tuberosity of BB and TA, respectively. Data was filtered (10–500 Hz) and sampled (1000 Hz) on a microSD card and later downloaded onto a PC for analysis.
Table 1 Technical specifications of lEMG (OT Bioelettronica, Torino, Italy) and Telemyo 900 (Noraxon USA Inc., Scottsdale, AZ).
Price Dimensions (mm) Weight (kg) Channels Bandwidth (Hz) Gain (V/V) Differential input impedance (X) Output range (V) Baseline Noise (mV RMS) Common Mode Rejection (dB) a
lEMG
Telemyo 900a
100€ ($130 USD) 57 L 36 W 16 H 0.025 1 10–500 980 >1011 0–3.3 <3 >95
$20,000 USD 165 L 70 W 25 H 0.453 8 10–500 2000 >106 0–5 <1 >85 dB @ 10–500 Hz >100 dB @ 50/60 Hz
Telemyo 900 includes transmitter and external computer connected receiver (specifications not shown).
Please cite this article in press as: Walters TJ et al. Validation of a portable EMG device to assess muscle activity during free-living situations. J Electromyogr Kinesiol (2013), http://dx.doi.org/10.1016/j.jelekin.2013.06.004
T.J. Walters et al. / Journal of Electromyography and Kinesiology xxx (2013) xxx–xxx
3
Fig. 1. Electrode arrangement. Electrodes for Telemyo 900 and lEMG were positioned over biceps brachii proximal and distal to the estimated location of the innervation zone, respectively.
Telemyo 900: Telemyo 900 was positioned in a pack around the waist and bipolar electrodes (10 10 mm, inter-electrode distance: 25 mm, Ag/AgCl; Vermed, Bellow Falls, VT) were positioned over BB and TA. Electrodes were secured to two leads of the 8-channel Telemyo 900 telemetry transmitter and a common ground electrode was positioned over the proximal ulna. Data was filtered (10–500 Hz) and sampled (2000 Hz) with a 16-bit ADC (Power 1401 and Spike2, Cambridge Electronic Design, UK) and stored on a computer. Electrode placement: Electrodes were positioned to avoid the innervation zone and tendon endings of BB and TA (Rainoldi et al., 2004; Saitou et al., 2000) and pilot testing involved using a 16-channel EMG array (EMGUSB2; OT Bioelettronica, Torino, IT) to confirm and guide electrode placement (Keenan et al., 2011). A counterbalanced design was used to control for electrode location in relation to the innervation zone and to minimize the influence of electrode placement on EMG amplitudes. Specifically, five subjects wore the lEMG concentric electrode proximal to the innervation zone (with Telemyo 900 electrodes distal to the innervation zone) and vice versa. Prior to electrode application, recording sites were shaven, abraded with paste, and cleaned. All EMG units, lead wires, and electrodes were wrapped with flexible bandages to reduce movement artifact. 2.3. Experimental procedure To compare the two EMG systems across a range of static forces, subjects performed at least two maximal voluntary contractions (MVCs), three submaximal contractions, and two 30-s static body-weight holds against gravity for each target muscle (Fig. 2A and B). Seated rest periods were also provided between contractions to determine baseline noise levels across EMG systems. MVCs for BB were performed against a handheld dynamometer (JAMAR;
Fig. 2. Representative EMGs detected from biceps brachii with simultaneous recordings from lEMG and Telemyo 900. EMGs from lEMG (A) and Telemyo 900 (B) are shown as subjects performed two maximal voluntary contractions (MVCs), three contractions at 25%, 50%, and 75% MVC, and while twice holding 90° of elbow flexion against gravity. EMGs from lEMG (C) and Telemyo 900 (D) are also shown during 20 min of free living.
Sammons Preston, Inc, Bolingbrook, Ill) positioned with a seatbelt restraint as the subjects sat upright on the edge of a training table with the elbow positioned in 90° of flexion. MVCs for TA were performed also against the JAMAR dynamometer as subjects were in a supine position on the training table, with the ankle at 0° dorsiflexion. After MVCs, subjects were asked to perform submaximal contractions of each target muscle at 25%, 50% and 75% of each muscle’s respective MVC force for at least five seconds. Target forces for each submaximal contraction were calculated using the highest force recorded from the dynamometer during the MVCs. Also, to evaluate the reliability of the two EMG systems from the start to the end of the experiment, EMGs during MVCs and submaximal contractions, as well as the baseline noise level, were assessed again at the end of the recording session. To compare the two EMG systems across a range of dynamic tasks, subjects first completed a battery of seven simulated ADLs, including upper and lower body tasks. Relatively light tasks were chosen to match the intensity of normal ADLs (Kern et al., 2001; Klein et al., 2010). Tasks included: (1) treadmill walking (0% grade, 2.0 mph), (2) incline treadmill walking (3% grade, 2.5 mph), (3) sitto-stand from a desk chair, (4) box carry (0.8 kg) from one table to
Please cite this article in press as: Walters TJ et al. Validation of a portable EMG device to assess muscle activity during free-living situations. J Electromyogr Kinesiol (2013), http://dx.doi.org/10.1016/j.jelekin.2013.06.004
4
T.J. Walters et al. / Journal of Electromyography and Kinesiology xxx (2013) xxx–xxx
another, (5) box step-ups using a 13 cm platform, (6) grooved pegboard test (Lafayette Instrument Company, Layfayette, IN), and (7) seated coffee-mug move (0.4 kg) from a tabletop onto and off of an elevated surface (19 cm). ADLs 3, 5, and 7 were completed as repetitive, forced-pace movements, with movement frequency set by a metronome at 45 BPM. Except for grooved pegboard, each ADL task lasted approximately 75 s with 60 s of seated rest between each task. Second, subjects completed over 60 min of simulated free-living time, in which the only instruction was to remain near the laboratory because the Telemyo 900 receiver needed to be within 300 yards of the transmitter to receive data (Fig. 2C and D).
during simulated upper body ADLs, within-subjects factors included the three upper body tasks (box carry, grooved pegboard, and seated coffee-mug move). For TA EMG amplitudes during simulated lower body ADLs, within-subjects factors included the four lower body tasks (treadmill walking, incline treadmill, sit-to-stand, and box step-ups). For EMG amplitudes during baseline noise within-subjects factors included time and muscle. For EMG amplitudes and durations during free-living within-subjects factors included analysis threshold (2%, 3%, and 4%) and muscle. Statistical significance was set at a = 0.05. 3. Results
2.4. Data analysis All data were analyzed using custom-designed software written in MATLAB (Mathworks, Natick, MA). As commonly done (Jakobi et al., 2008; Klein et al., 2010; Ostensvik et al., 2009), EMG signals were analyzed visually to identify artifacts and none were clearly evident. Although high-pass filters for both EMG systems were set at 10 Hz, EMGs were subsequently high-pass filtered in MATLAB using a 4th order, zero-lag, Butterworth filter with a cutoff frequency of 20 Hz to further reduce movement artifact and noise (De Luca et al., 2010). In order to estimate EMG amplitude, the EMG signals were full-wave rectified and smoothed using a non-overlapping 10 ms window. MVCs: As in Klein et al. (2010), EMG amplitudes during the MVCs at the start and the end of the experiment were calculated from the largest mean amplitude averaged over 200 ms during the MVC period (EMGmax). All EMG data were normalized as a percentage of each participant’s EMGmax. Baseline Noise: Baseline noise was computed over a 30-s segment of EMG during periods of inactivity at the start and end of the experiment. Submaximal Contractions: EMG amplitude during the three submaximal contractions was calculated over a 5-s segment visually selected by the experimenter to avoid the start and end of each contraction). Simulated ADLs: EMG amplitude during the seven simulated ADLs was calculated using a 30-s segment of EMG visually selected to avoid the start and end of the tasks. Free Living: Periods of significant EMG activity during free living were determined as in Klein et al. (2010). Since analysis thresholds can affect variables derived from the EMG signal (Klein et al., 2010), three thresholds (2%, 3%, and 4% EMGmax) were used to quantify periods of significant EMG activity during free living. Significant muscle activity was defined as a 10-ms segment of EMG data that exceeded the predetermined threshold. To estimate EMG amplitudes, EMG segments that exceeded the threshold were averaged and normalized to EMGmax. EMG durations during the free-living period were estimated as the product of the number of segments above threshold and the segment duration of 10 ms, and then expressed as a percentage of the total free-living time, which varied across subjects.
As shown in Fig. 2, EMGs from both systems were qualitatively similar for the different tasks. Quantitative similarities and differences between the two systems are highlighted below. 3.1. MVCs There was no significant main effect of recording time as absolute EMG amplitudes for MVCs at the start and end of the experimental protocol were statistically similar (0.47 ± 0.15 mV vs. 0.40 ± 0.13 mV, respectively; p = .088; Fig. 3). As expected, absolute EMG amplitudes during MVCs for the Telemyo 900 were greater compared with lEMG that used the more selective concentric electrodes (0.54 ± 0.17 mV vs. 0.37 ± 0.18 mV, respectively; p = .036). In addition, EMG amplitudes were increased for BB compared with TA (0.57 ± 0.24 mV vs. 0.33 ± 0.08 mV, respectively; p = .012). No interaction was significant (p > .116). These results indicate that the performance of the two EMG systems was stable for the two muscles (BB and TA) examined at the start and end of the experimental protocol. 3.2. Baseline noise There was no significant main effect of baseline noise (normalized to EMGmax) between the lEMG and the Telemyo 900 systems (0.27 ± 0.17% EMGmax vs. 0.19 ± 0.08% EMGmax, respectively; p = .174; Fig. 4). However, there was a significant interaction between EMG amplifier and muscle (p = 0.033) with greater baseline noise in TA when detected with lEMG compared with Telemyo 900 (p = 0.045), though the magnitude of the difference was rela-
2.5. Statistical analysis Within-subjects repeated measures ANOVA was used to test for differences in EMG parameters between the lEMG and Telemyo 900 units. For all analyses, within-subjects factors included the EMG system (lEMG or Telemyo 900) with additional factors and dependent variables as described below. For EMG amplitudes during MVCs, within-subjects factors included MVC time (start/end) and muscle (BB/TA). For EMG amplitudes during submaximal force contractions, within-subjects factors included contraction intensity (25%, 50%, and 75% MVC) and muscle. For BB EMG amplitudes
Fig. 3. Absolute EMG amplitudes during MVC. There were no differences in absolute EMG amplitudes (mV) for MVCs at the start and end of the experiment (p = .088). Absolute EMG amplitudes were greater while using the Telemyo 900 compared with lEMG, likely because lEMG used more selective concentric electrodes. In addition, absolute EMG amplitudes were greater for biceps brachii (BB) compared with tibialis anterior (TA). Values are means ± SE. ap = .036 vs. lEMG; bp = .012 vs. TA.
Please cite this article in press as: Walters TJ et al. Validation of a portable EMG device to assess muscle activity during free-living situations. J Electromyogr Kinesiol (2013), http://dx.doi.org/10.1016/j.jelekin.2013.06.004
T.J. Walters et al. / Journal of Electromyography and Kinesiology xxx (2013) xxx–xxx
Fig. 4. Baseline noise from lEMG and Telemyo 900. There were no differences in baseline noise, normalized as a percent of maximal EMG, between the two EMG systems (p = .174). However, there was an EMG amplifier by muscle interaction (p = 0.033), with greater baseline noise in tibialis anterior (TA) when detected with lEMG. In addition, there was an EMG amplifier by time interaction (p = 0.003) with greater baseline noise at the start of the experiment when detected with lEMG. Nonetheless, baseline noise was always below 0.5% maximum EMG (EMGmax). Values are means ± SE. ap = 0.045 vs. baseline noise detected in TA with Telemyo 900; bp = 0.043 vs. baseline noise detected at the start of the experiment with Telemyo 900.
5
Fig. 5. EMG amplitudes during 25%, 50%, and 75% MVC contractions. There were no significant differences in EMG amplitudes between the lEMG and Telemyo 900 systems (p = .936), and all other main effects and interactions were not significant (p > .094). EMG amplitudes detected simultaneously from the two systems were plotted for both muscles. As expected there was variability in individual values, however the slope of the best-fit line was 1.00 and the R2 value was 0.82. Also note that subjects performed the task without visual feedback and frequently underestimated the amount of force necessary for the task, contributing to the large range of forces produced.
tively small at 0.15% EMGmax. In addition, we found a significant interaction between EMG amplifier and time (p = 0.003) with greater baseline noise at the start of the experiment when detected with lEMG compared with Telemyo 900 (p = 0.043), though the magnitude of the difference was also small and 0.13% EMGmax. No other interaction was significant (p > .244). Although there were significant differences between the two EMG systems in terms of baseline noise, the differences were below 0.15% EMGmax and baseline noise did not exceed 0.5% EMGmax.
3.3. Submaximal contractions There were no significant differences for the three submaximal force contractions between the lEMG and the Telemyo 900 systems (p = .936), and all other main effects and interactions were not significant (p > .094). Note that subjects performed the submaximal contractions without visual feedback, thus there was large variability in the submaximal forces actually produced, with subjects generally producing force magnitudes less than the target force level. Fig. 5 is a scatter plot that shows the normalized EMG amplitudes from both EMG systems during the three submaximal contractions for all subjects. Although there was variability in individual values, as commonly expected with EMG and performing the task without feedback, the slope of the line representing the best linear fit of the data is 1.00 (Fig. 5), and the R2 value is 0.82, suggesting that the EMG systems recorded muscle activity similarly across varying magnitudes of muscle contraction.
3.4. Simulated activities of daily living There were no significant differences for EMGs from BB during the ‘‘upper’’ body simulated ADLs between lEMG and Telemyo 900 systems (4.50 ± 2.46% EMGmax vs. 5.21 ± 3.05% EMGmax; p = .488, respectively; Fig. 6A), nor for EMGs from TA during the ‘‘lower’’ body ADLs (13.44 ± 7.12% EMGmax vs. 9.81 ± 2.45% EMGmax; p = .199, respectively; Fig. 6B). There was a significant main effect for task in TA (p = .037) but not for BB (p = .483; Fig. 6). EMG amplifier by task interactions for BB and TA were not significant
Fig. 6. EMG amplitudes during simulated activities of daily living. There were no significant differences between lEMG and Telemyo 900 systems for (A) EMG amplitudes from biceps brachii (BB) during the ‘‘upper’’ body tasks (p = .488), nor for (B) EMG amplitudes from tibialis anterior (TA) during the ‘‘lower’’ body tasks (p = .199). All interactions including the EMG amplifier were not significant (p > .127) indicating the EMG systems performed similarly across tasks.
Please cite this article in press as: Walters TJ et al. Validation of a portable EMG device to assess muscle activity during free-living situations. J Electromyogr Kinesiol (2013), http://dx.doi.org/10.1016/j.jelekin.2013.06.004
6
T.J. Walters et al. / Journal of Electromyography and Kinesiology xxx (2013) xxx–xxx
(p = .263, p = .127, respectively), which indicated that the EMG systems performed similarly during each individual task. 3.5. Free living There were no significant differences between the lEMG and the Telemyo 900 systems for normalized EMG amplitudes (p = .667) nor for EMG durations (p = .581) during the simulated free-living period (Fig. 7). There was a significant main effect for threshold in EMG amplitudes and durations during free living (p < .001, p = .001, respectively). All interactions for normalized EMG amplitudes and EMG durations were not significant (p > .050). As the use of Telemyo 900 restricted recording EMGs within 300 yards of the laboratory, we tested the feasibility of using lEMG for a longer wear time during a free-living period outside the laboratory. Fig. 8 depicts representative EMG data recorded from BB and TA and accelerations from a hip-worn accelerometer (GT3X+; Actigraph, Pensacola, FL). Although a comparison to Telemyo 900 was not possible, lEMG successfully recorded EMGs during the longer wear time. 4. Discussion This is the first study to systematically validate a low-cost EMG data logging system designed specifically for free-living situations.
Fig. 8. Representative data during a longer free-living period performed outside of the laboratory. To test the feasibility of recording EMGs for longer periods of time with lEMG, EMGs were detected from (A) biceps brachii and (B) tibialis anterior during a 7 h free-living period. In addition, average counts/min from a (C) hipworn accelerometer was also computed. EMG and accelerometry data was collected simultaneously and is shown in 1-s epochs.
Results from the current study indicate that the portable, low-cost lEMG and the laboratory-based Telemyo 900 system performed similarly in the current study, supporting the use of lEMG for recording EMGs during free living situations. Strengths and limitations of lEMG are discussed below. 4.1. Strengths of lEMG First, similarities across the two EMG systems in the current study in terms of EMG parameters during submaximal contractions, simulated ADLs, and free-living support the use of lEMG. As clearly represented in Fig. 1, the EMGs from the two systems are nearly identical and there was no clear artifact while recording from either system. Second, similar to using accelerometers outside of the laboratory to track physical activity (Swartz et al., 2000; Troiano et al., 2008), the low-cost and small size of lEMG allows deployment of multiple units simultaneously from the laboratory or a remote location. In addition, lEMG can be deployed without a large financial risk as is common when sending higher-cost, laboratory-based EMG systems into free-living situations. The ease of use and deployment of lEMG should expand the application of long-term EMG within the fields of aging, ergonomics, and clinical disorders. Third, the design of the concentric electrode for the lEMG system enhances its practicality in free-living situations. Advantages of the concentric electrode over traditional bipolar recordings include: easy application of the electrode as its orientation with respect to muscle fiber angle or changes in fiber angle will have minimal influence on the EMG signal (Farina and Cescon, 2001; Rainoldi et al., 2004); higher spatial selectivity with respect to traditional detection systems (Farina and Cescon, 2001) and thereby reducing the potential confound of crosstalk (De Luca and Merletti, 1988; Farina et al., 2002; Solomonow et al., 1994); and the concentric surface electrode used with lEMG has a large adhesive surface with a short cable length, which reduces the problem of electrode adherence during long-term recordings and potential artifacts due to cable movement.
Fig. 7. Free-living EMG amplitudes and durations. EMG amplitudes and durations were estimated using three different analysis thresholds (2%, 3%, and 4% EMGmax; see Section 2). There were no significant differences between the two EMG systems for (A) EMG amplitudes (p = .667), nor for (B) EMG durations (p = .581). All interactions for EMG amplitudes and EMG durations were not significant (p > .050, p > .073, respectively).
4.2. Limitations of lEMG First, baseline noise was not identical across both EMG systems. Specifically, there was greater baseline noise in TA when detected
Please cite this article in press as: Walters TJ et al. Validation of a portable EMG device to assess muscle activity during free-living situations. J Electromyogr Kinesiol (2013), http://dx.doi.org/10.1016/j.jelekin.2013.06.004
T.J. Walters et al. / Journal of Electromyography and Kinesiology xxx (2013) xxx–xxx
with lEMG compared with Telemyo 900 (p = 0.045) and greater baseline noise at the start of the experiment when detected with lEMG compared with Telemyo 900 (p = 0.043; Fig. 4). It is not clear why these differences exist; however, the magnitude of the differences across systems was small (<0.15% EMGmax) and baseline noise for both systems was less than 0.5% EMGmax. Nonetheless, although amplitude thresholds at 2% EMGmax are most commonly used in long-term EMG studies (Kern et al., 2001; Klein et al., 2010; Theou et al., 2010), some studies use lower thresholds at 0.5% EMGmax to investigate how sustained low-activity muscle activation relates to pain (Mork and Westgaard, 2005; Ostensvik et al., 2009). Our study suggests caution in deploying lEMG with such low-amplitude thresholds, at least for some muscles like TA where baseline noise can approach 0.5% EMGmax (Fig. 4). Second, unlike all laboratory-based EMG systems, instantaneous visual feedback of EMG is not possible with lEMG. Thus, to visually inspect EMG signals from the lEMG, the experimenter must eject the microSD card and load it onto a computer with the OT Biolab software. Third, since lEMG exhibits a limited output range compared to laboratory-based systems (±3.3 V vs. ±5 V, respectively), and there is no adjustable gain with lEMG, saturation of the EMG signal is more likely. For example, in early piloting of lEMG we tested vastus medialis but found the signal saturated during some MVC trials. However, signal saturation was not an issue for BB or TA. This issue could limit the usefulness of lEMG when recording EMGs from other muscles. Nonetheless, normalization of EMG amplitude to submaximal contractions is likely a viable solution (Yang and Winter, 1983) as free-living muscle activity is commonly performed at low intensity levels (Kern et al., 2001; Klein et al., 2010). In summary, this was the first study to validate a low-cost EMG data logging system specifically designed for use outside of the laboratory. Comparison between a low-cost system and the laboratory-based system revealed similarities in several outcome variables, which suggests that the lEMG system is a valid approach to recording EMGs during free-living situations. Though the lEMG system exhibits certain constraints compared to the Telemyo 900 system, specifically related to baseline noise, the design of the lEMG system makes it a viable approach for large-scale investigation of long-term, free-living neuromuscular function across different subject populations (e.g., healthy, elderly, and stroke). Similar to the miniaturization and common deployment of accelerometers to measure physical activity, low-cost, portable EMG systems specifically designed for free-living situations will allow researchers to better assess the functional use of muscles during activities of daily living.
7
Kern DS, Semmler JG, Enoka RM. Long-term activity in upper- and lower-limb muscles of humans. J Appl Physiol 2001;91:2224–32. Klein CS, Peterson LB, Ferrell S, Thomas CK. Sensitivity of 24-h EMG duration and intensity in the human vastus lateralis muscle to threshold changes. J Appl Physiol 2010;108:655–61. Monster AW, Chan H, O’Connor D. Activity patterns of human skeletal muscles: relation to muscle fiber type composition. Science 1978;200:314–7. Mork PJ, Westgaard RH. Long-term electromyographic activity in upper trapezius and low back muscles of women with moderate physical activity. J Appl Physiol 2005;99:570–8. Ostensvik T, Veiersted KB, Nilsen P. A method to quantify frequency and duration of sustained low-level muscle activity as a risk factor for musculoskeletal discomfort. J Electromyogr Kinesiol 2009;19:283–94. Rainoldi A, Melchiorri G, Caruso I. A method for positioning electrodes during surface EMG recordings in lower limb muscles. J Neurosci Methods. 2004;134:37–43. Saitou K, Masuda T, Michikami D, Kojima R, Okada M. Innervation zones of the upper and lower limb muscles estimated by using multichannel surface EMG. J Hum Ergol (Tokyo) 2000;29:35–52. Solomonow M, Baratta R, Zhou B, Lu Y, Zhu M, Acierno S. Surface and wire EMG crosstalk in neighbouring muscles. J Electromyogr Kinesiol 1994;4:131–42. Swartz AM, Strath SJ, Bassett Jr DR, O’Brien WL, King GA, Ainsworth BE. Estimation of energy expenditure using CSA accelerometers at hip and wrist sites. Med Sci Sports Exerc 2000;32:S450–6. Theou O, Jones GR, Vandervoort AA, Jakobi JM. Daily muscle activity and quiescence in non-frail, pre-frail, and frail older women. Exp Gerontol 2010;45:909–17. Troiano RP, Berrigan D, Dodd KW, Masse LC, Tilert T, McDowell M. Physical activity in the United States measured by accelerometer. Med Sci Sports Exerc 2008;40:181–8. Yang JF, Winter DA. Electromyography reliability in maximal and submaximal isometric contractions. Archiv Phys Med Rehab 1983;64:417–20.
Tygh J. Walters received his B.A. in Exercise Physiology and Human Performance from Ripon College in Ripon, WI in 2009 prior to completing his M.S. degree in Kinesiology from the University of WisconsinMilwaukee in 2012. His research interests focus on physical function and long-term electromyography in free-living situations. He is currently an assistant at an interventional pain clinic in Norfolk, NE.
Kayla Kaschinske received her B.S. in Kinesiology at the University of Wisconsin – Milwaukee in 2011. She is currently a Doctor of Physical Therapy student at the University of Wisconsin–Madison and expects to graduate in May of 2014.
Acknowledgement UW-Milwaukee Research Growth Initiative supported the study but had no other involvement. References De Luca CJ, Gilmore LD, Kuznetsov M, Roy SH. Filtering the surface EMG signal: movement artifact and baseline noise contamination. J Biomech 2010;43:1573–9. De Luca CJ, Merletti R. Surface myoelectric signal cross-talk among muscles of the leg. Electroencephalogr Clin Neurophysiol 1988;69:568–75. Farina D, Cescon C. Concentric-ring electrode systems for noninvasive detection of single motor unit activity. IEEE Trans Biomed Eng 2001;48:1326–34. Farina D, Merletti R, Indino B, Nazzaro M, Pozzo M. Surface EMG crosstalk between knee extensor muscles: experimental and model results. Muscle Nerve 2002;26:681–95. Jakobi JM, Edwards DL, Connelly DM. Utility of portable electromyography for quantifying muscle activity during daily use. Gerontology 2008;54:324–31. Keenan KG, Collins JD, Massey WV, Walters TJ, Gruszka HD. Coherence between surface electromyograms is influenced by electrode placement in hand muscles. J Neurosci Methods 2011;195:10–4.
Scott J. Strath received his B.Ed. in Physical Education (1996) from Sheffield Hallam University, his M.S. in Exercise Science (1998) from Ball State University, and his Ph.D. in Exercise Science from the University of Tennessee (2001). He then completed post-doctoral training in the Department of Physical Medicine and Rehabilitation at the University of Kentucky and the University of Michigan. He is currently an Associate Professor in the Department of Kinesiology at the University of Wisconsin-Milwaukee. His research interests revolve around physical activity and public health. Specific interests include the relationship between physical activity and cardiovascular health; physical activity promotional strategies; physical activity assessment; physical activity epidemiology; community and national physical activity patterns; and environmental determinants of physical activity behavior.
Please cite this article in press as: Walters TJ et al. Validation of a portable EMG device to assess muscle activity during free-living situations. J Electromyogr Kinesiol (2013), http://dx.doi.org/10.1016/j.jelekin.2013.06.004
8
T.J. Walters et al. / Journal of Electromyography and Kinesiology xxx (2013) xxx–xxx
Ann M. Swartz received her B.S. in Kinesiology (1995) from the University of Wisconsin-Madison, and M.S. in Exercise Science (1998) from Ball State University, and her Ph.D. in Exercise Science (2001) from the University of Tennessee. She then completed post-doctoral training in the Department of Physical Medicine and Rehabilitation at the University of Kentucky and the University of Michigan. She is currently an Associate Professor in the Department of Kinesiology at the University of Wisconsin-Milwaukee. Her research interests center on obesity; in particular, the health benefits of physical activity for overweight and obese individuals including 1) relationships between physical activity, health and obesity level, 2) novel translational physical activity interventions to improve health for overweight and obese individuals, and 3) the accurate assessment of physical activity behavior.
Kevin G. Keenan received his B.S. (1990) and M.S. (1998) degrees in Kinesiology from the University of Michigan. He subsequently worked as the Head Strength and Conditioning Coach at Providence College prior to receiving his Ph.D. in Integrative Physiology from the University of Colorado-Boulder (2005) and completing post-doctoral training (2005–2008) at Cornell University in the Sibley School of Mechanical and Aerospace Engineering. He is currently an Assistant Professor in the Department of Kinesiology at the University of Wisconsin-Milwaukee. His research interests focus on understanding the interplay between motor function, physical activity, and health using high-density surface EMG arrays, long-term (>6 h) surface EMG recordings, indwelling EMG recordings, and computational modeling.
Please cite this article in press as: Walters TJ et al. Validation of a portable EMG device to assess muscle activity during free-living situations. J Electromyogr Kinesiol (2013), http://dx.doi.org/10.1016/j.jelekin.2013.06.004
Contents lists available at SciVerse ScienceDirect
Journal of Electromyography and Kinesiology journal homepage: www.elsevier.com/locate/jelekin
Validation of a portable EMG device to assess muscle activity during free-living situations T.J. Walters, K.A. Kaschinske, S.J. Strath, A.M. Swartz, K.G. Keenan ⇑ University of Wisconsin-Milwaukee, Department of Kinesiology, Milwaukee, WI, USA
a r t i c l e
i n f o
Article history: Received 14 November 2012 Received in revised form 9 May 2013 Accepted 8 June 2013 Available online xxxx Keywords: Activities of daily living Movement Motor activity Muscle strength
a b s t r a c t Portable amplifiers that record electromyograms (EMGs) for longer than four hours are commonly priced over $20,000 USD. This cost, and the technical challenges associated with recording EMGs during freeliving situations, typically restrict EMG use to laboratory settings. A low-cost system (lEMG; OT Bioelecttronica, 100€), using specialized concentric bipolar electrodes, has been developed specifically for free-living situations. The purpose of this study was to validate the lEMG system by comparing EMGs from lEMG with a laboratory-based alternative (Telemyo 900; Noraxon USA, Inc.). Surface EMGs from biceps brachii (BB) and tibialis anterior (TA) of ten subjects were recorded simultaneously with both systems as subjects performed maximal voluntary contractions (MVCs), submaximal contractions at 25%, 50%, and 75% MVC, seven simulated activities of daily living (ADLs), and >60 min of simulated free-living inside the laboratory. In general, EMG parameters (e.g., average full-wave rectified EMG amplitude) derived from both systems were not significantly different for all outcome variables, except there were small differences across systems in baseline noise and absolute EMG amplitudes during MVCs. These results suggest that lEMG is a valid approach to the long-term recording of EMG. Ó 2013 Published by Elsevier Ltd.
1. Introduction Accurate measurement of neuromuscular function in free-living situations is critical to assess the functional use of muscles during activities of daily living (ADLs) and has been used to investigate work-related pain (Ostensvik et al., 2009), advancing age (Theou et al., 2010), and clinical disorder (Jakobi et al., 2008). Although movement during free-living situations is most commonly studied using questionnaires, pedometers, and accelerometers, only electromyography (EMG) can indicate the neuromuscular strategy underlying movement during daily activities. Unfortunately, there are a number of technical challenges associated with using laboratory-based, portable EMG systems during free living. These challenges include spurious artifacts and noise in long-term EMG recordings that arise from movement of the EMG system and cabling when used across different environments (Jakobi et al., 2008; Klein et al., 2010). These technical challenges, as well as the high cost of conventional EMG systems, typically restrict their use to laboratory settings and long-term EMG recordings from human muscles are still relatively scarce (e.g., Jakobi et al., 2008; Kern et al., 2001; Klein et al., 2010; Monster et al., 1978; Mork and
⇑ Corresponding author. Address: The University of Wisconsin-Milwaukee, 2400 E. Hartford Ave., Milwaukee, WI 53211, USA. Tel.: +1 (414) 229 2336; fax: +1 (414) 229 2619. E-mail address: [email protected] (K.G. Keenan).
Westgaard, 2005; Ostensvik et al., 2009). Alternatively, low-cost EMG data logging systems have been developed specifically for use during free living, though the validity of these low-cost EMG systems has not been examined. One system developed specifically for recording EMGs during free living is lEMG (OT Bioelettronica, Torino, Italy). In contrast to typical laboratory-based EMG systems, lEMG is inexpensive (100€), small (57 mm 36 mm 16 mm), and lightweight (25 g). Similar to hip-worn accelerometers commonly used to estimate energy expenditure and track physical activity outside the laboratory (Swartz et al., 2000; Troiano et al., 2008), the low-profile lEMG is easily positioned on the subject and can be deployed from any convenient location to collect EMG data continuously during free-living situations for up to nine hours. Since laboratory-based EMG systems typically cost over $20,000 USD, the relative low-cost and specific design of lEMG (as highlighted below) may make it a practical solution to larger-scale, out-of-laboratory data collections. However, there are substantial differences between lEMG and laboratory-based systems that warrant validation of lEMG. Potential limitations of the lEMG system include: one amplification gain (980 V/V), a limited output range of 3.3 V, and an A/D resolution of 8 bit. Thus, the resolution of lEMG is limited compared with laboratory-based systems and more susceptible to signal saturation which could decrease its usefulness. Moreover, it is unclear how lEMG is influenced by different sources of noise and signal artifact
1050-6411/$ - see front matter Ó 2013 Published by Elsevier Ltd. http://dx.doi.org/10.1016/j.jelekin.2013.06.004
Please cite this article in press as: Walters TJ et al. Validation of a portable EMG device to assess muscle activity during free-living situations. J Electromyogr Kinesiol (2013), http://dx.doi.org/10.1016/j.jelekin.2013.06.004
2
T.J. Walters et al. / Journal of Electromyography and Kinesiology xxx (2013) xxx–xxx
that can be common in free-living situations (De Luca et al., 2010; Klein et al., 2010). In contrast to these possible limitations, lEMG has features that may enhance its practicality for use in free-living situations compared with laboratory-based systems. Specifically, lEMG uses concentric detection bipolar lead electrodes which are isotropic, or invariant to rotations of the electrode with respect to the underlying muscle fiber orientation (Farina and Cescon, 2001). Consequently, the concentric electrodes can be placed without consideration of muscle fiber orientation or changes in muscle fiber orientation that typically occur during anisometric contractions (Farina and Cescon, 2001). Furthermore, concentric electrodes are more selective and thus less susceptible to the problem of crosstalk (Farina and Cescon, 2001). In addition, the concentric electrode developed by OT Bioelettronica has a low-profile design and large adhesive surface that is unlikely to be dislodged after hours of wear, as well as short cable lengths that may minimize artifacts. Nonetheless, although the advantages of surface concentric electrodes have been well described theoretically using numerical models (Farina and Cescon, 2001), no experimental studies have yet used the surface concentric electrodes. The purpose of this study was to validate lEMG by comparing EMGs from lEMG with simultaneous recordings from Noraxon Telemyo 900. The study involved having subjects perform a range of isometric forces, seven simulated ADLs and a short (>1 h) simulated free-living condition. EMGs were collected simultaneously with both EMG systems from biceps brachii (BB) and tibialis anterior (TA) and EMG parameters were compared across systems.
2. Methods 2.1. Subjects Ten healthy subjects (5 males, 24.5 ± 7.4 yrs (mean ± SD); range: 19–44 yrs old) participated in the study. All subjects
reported to be healthy and free of known orthopedic or neuromuscular disorders. All participants read and signed an informed consent document approved by the University Institutional Review Board prior to participation. 2.2. Setup Table 1 highlights specifications for both lEMG and the laboratory-based system (Telemyo 900, Noraxon USA, Inc.) used for comparison in the current study. EMGs were collected simultaneously with both EMG systems with electrodes positioned on the skin overlying biceps brachii (BB) and tibialis anterior (TA). BB and TA were selected as they are relatively long muscles with clearly identified innervation zones and tendon endings (Rainoldi et al., 2004; Saitou et al., 2000); thus, placing electrodes from both systems on either side of the innervation zone was possible (Fig. 1). In addition, both muscles have relatively low pennation angles and are less influenced by crosstalk because of their relatively isolated anatomical positions, thus eliminating advantages of the lEMG system and allowing the Noraxon Telemyo 900 to provide a more realistic criterion for comparison to lEMG.
lEMG: Each subject wore two lEMGs with concentric electrodes. Concentric electrodes consisted of Ag/AgCl with a circular point electrode (1.6 cm diameter) and surrounding external circular ring (outside and inside diameter of 5 cm and 4 cm, respectively) and were placed on the skin overlying BB and TA on the right side of the body. lEMG units were secured and grounded with square adhesive electrodes (10 10 mm; Vermed, Bellow Falls, VT) over the acromio-clavicular joint and over the tibial tuberosity of BB and TA, respectively. Data was filtered (10–500 Hz) and sampled (1000 Hz) on a microSD card and later downloaded onto a PC for analysis.
Table 1 Technical specifications of lEMG (OT Bioelettronica, Torino, Italy) and Telemyo 900 (Noraxon USA Inc., Scottsdale, AZ).
Price Dimensions (mm) Weight (kg) Channels Bandwidth (Hz) Gain (V/V) Differential input impedance (X) Output range (V) Baseline Noise (mV RMS) Common Mode Rejection (dB) a
lEMG
Telemyo 900a
100€ ($130 USD) 57 L 36 W 16 H 0.025 1 10–500 980 >1011 0–3.3 <3 >95
$20,000 USD 165 L 70 W 25 H 0.453 8 10–500 2000 >106 0–5 <1 >85 dB @ 10–500 Hz >100 dB @ 50/60 Hz
Telemyo 900 includes transmitter and external computer connected receiver (specifications not shown).
Please cite this article in press as: Walters TJ et al. Validation of a portable EMG device to assess muscle activity during free-living situations. J Electromyogr Kinesiol (2013), http://dx.doi.org/10.1016/j.jelekin.2013.06.004
T.J. Walters et al. / Journal of Electromyography and Kinesiology xxx (2013) xxx–xxx
3
Fig. 1. Electrode arrangement. Electrodes for Telemyo 900 and lEMG were positioned over biceps brachii proximal and distal to the estimated location of the innervation zone, respectively.
Telemyo 900: Telemyo 900 was positioned in a pack around the waist and bipolar electrodes (10 10 mm, inter-electrode distance: 25 mm, Ag/AgCl; Vermed, Bellow Falls, VT) were positioned over BB and TA. Electrodes were secured to two leads of the 8-channel Telemyo 900 telemetry transmitter and a common ground electrode was positioned over the proximal ulna. Data was filtered (10–500 Hz) and sampled (2000 Hz) with a 16-bit ADC (Power 1401 and Spike2, Cambridge Electronic Design, UK) and stored on a computer. Electrode placement: Electrodes were positioned to avoid the innervation zone and tendon endings of BB and TA (Rainoldi et al., 2004; Saitou et al., 2000) and pilot testing involved using a 16-channel EMG array (EMGUSB2; OT Bioelettronica, Torino, IT) to confirm and guide electrode placement (Keenan et al., 2011). A counterbalanced design was used to control for electrode location in relation to the innervation zone and to minimize the influence of electrode placement on EMG amplitudes. Specifically, five subjects wore the lEMG concentric electrode proximal to the innervation zone (with Telemyo 900 electrodes distal to the innervation zone) and vice versa. Prior to electrode application, recording sites were shaven, abraded with paste, and cleaned. All EMG units, lead wires, and electrodes were wrapped with flexible bandages to reduce movement artifact. 2.3. Experimental procedure To compare the two EMG systems across a range of static forces, subjects performed at least two maximal voluntary contractions (MVCs), three submaximal contractions, and two 30-s static body-weight holds against gravity for each target muscle (Fig. 2A and B). Seated rest periods were also provided between contractions to determine baseline noise levels across EMG systems. MVCs for BB were performed against a handheld dynamometer (JAMAR;
Fig. 2. Representative EMGs detected from biceps brachii with simultaneous recordings from lEMG and Telemyo 900. EMGs from lEMG (A) and Telemyo 900 (B) are shown as subjects performed two maximal voluntary contractions (MVCs), three contractions at 25%, 50%, and 75% MVC, and while twice holding 90° of elbow flexion against gravity. EMGs from lEMG (C) and Telemyo 900 (D) are also shown during 20 min of free living.
Sammons Preston, Inc, Bolingbrook, Ill) positioned with a seatbelt restraint as the subjects sat upright on the edge of a training table with the elbow positioned in 90° of flexion. MVCs for TA were performed also against the JAMAR dynamometer as subjects were in a supine position on the training table, with the ankle at 0° dorsiflexion. After MVCs, subjects were asked to perform submaximal contractions of each target muscle at 25%, 50% and 75% of each muscle’s respective MVC force for at least five seconds. Target forces for each submaximal contraction were calculated using the highest force recorded from the dynamometer during the MVCs. Also, to evaluate the reliability of the two EMG systems from the start to the end of the experiment, EMGs during MVCs and submaximal contractions, as well as the baseline noise level, were assessed again at the end of the recording session. To compare the two EMG systems across a range of dynamic tasks, subjects first completed a battery of seven simulated ADLs, including upper and lower body tasks. Relatively light tasks were chosen to match the intensity of normal ADLs (Kern et al., 2001; Klein et al., 2010). Tasks included: (1) treadmill walking (0% grade, 2.0 mph), (2) incline treadmill walking (3% grade, 2.5 mph), (3) sitto-stand from a desk chair, (4) box carry (0.8 kg) from one table to
Please cite this article in press as: Walters TJ et al. Validation of a portable EMG device to assess muscle activity during free-living situations. J Electromyogr Kinesiol (2013), http://dx.doi.org/10.1016/j.jelekin.2013.06.004
4
T.J. Walters et al. / Journal of Electromyography and Kinesiology xxx (2013) xxx–xxx
another, (5) box step-ups using a 13 cm platform, (6) grooved pegboard test (Lafayette Instrument Company, Layfayette, IN), and (7) seated coffee-mug move (0.4 kg) from a tabletop onto and off of an elevated surface (19 cm). ADLs 3, 5, and 7 were completed as repetitive, forced-pace movements, with movement frequency set by a metronome at 45 BPM. Except for grooved pegboard, each ADL task lasted approximately 75 s with 60 s of seated rest between each task. Second, subjects completed over 60 min of simulated free-living time, in which the only instruction was to remain near the laboratory because the Telemyo 900 receiver needed to be within 300 yards of the transmitter to receive data (Fig. 2C and D).
during simulated upper body ADLs, within-subjects factors included the three upper body tasks (box carry, grooved pegboard, and seated coffee-mug move). For TA EMG amplitudes during simulated lower body ADLs, within-subjects factors included the four lower body tasks (treadmill walking, incline treadmill, sit-to-stand, and box step-ups). For EMG amplitudes during baseline noise within-subjects factors included time and muscle. For EMG amplitudes and durations during free-living within-subjects factors included analysis threshold (2%, 3%, and 4%) and muscle. Statistical significance was set at a = 0.05. 3. Results
2.4. Data analysis All data were analyzed using custom-designed software written in MATLAB (Mathworks, Natick, MA). As commonly done (Jakobi et al., 2008; Klein et al., 2010; Ostensvik et al., 2009), EMG signals were analyzed visually to identify artifacts and none were clearly evident. Although high-pass filters for both EMG systems were set at 10 Hz, EMGs were subsequently high-pass filtered in MATLAB using a 4th order, zero-lag, Butterworth filter with a cutoff frequency of 20 Hz to further reduce movement artifact and noise (De Luca et al., 2010). In order to estimate EMG amplitude, the EMG signals were full-wave rectified and smoothed using a non-overlapping 10 ms window. MVCs: As in Klein et al. (2010), EMG amplitudes during the MVCs at the start and the end of the experiment were calculated from the largest mean amplitude averaged over 200 ms during the MVC period (EMGmax). All EMG data were normalized as a percentage of each participant’s EMGmax. Baseline Noise: Baseline noise was computed over a 30-s segment of EMG during periods of inactivity at the start and end of the experiment. Submaximal Contractions: EMG amplitude during the three submaximal contractions was calculated over a 5-s segment visually selected by the experimenter to avoid the start and end of each contraction). Simulated ADLs: EMG amplitude during the seven simulated ADLs was calculated using a 30-s segment of EMG visually selected to avoid the start and end of the tasks. Free Living: Periods of significant EMG activity during free living were determined as in Klein et al. (2010). Since analysis thresholds can affect variables derived from the EMG signal (Klein et al., 2010), three thresholds (2%, 3%, and 4% EMGmax) were used to quantify periods of significant EMG activity during free living. Significant muscle activity was defined as a 10-ms segment of EMG data that exceeded the predetermined threshold. To estimate EMG amplitudes, EMG segments that exceeded the threshold were averaged and normalized to EMGmax. EMG durations during the free-living period were estimated as the product of the number of segments above threshold and the segment duration of 10 ms, and then expressed as a percentage of the total free-living time, which varied across subjects.
As shown in Fig. 2, EMGs from both systems were qualitatively similar for the different tasks. Quantitative similarities and differences between the two systems are highlighted below. 3.1. MVCs There was no significant main effect of recording time as absolute EMG amplitudes for MVCs at the start and end of the experimental protocol were statistically similar (0.47 ± 0.15 mV vs. 0.40 ± 0.13 mV, respectively; p = .088; Fig. 3). As expected, absolute EMG amplitudes during MVCs for the Telemyo 900 were greater compared with lEMG that used the more selective concentric electrodes (0.54 ± 0.17 mV vs. 0.37 ± 0.18 mV, respectively; p = .036). In addition, EMG amplitudes were increased for BB compared with TA (0.57 ± 0.24 mV vs. 0.33 ± 0.08 mV, respectively; p = .012). No interaction was significant (p > .116). These results indicate that the performance of the two EMG systems was stable for the two muscles (BB and TA) examined at the start and end of the experimental protocol. 3.2. Baseline noise There was no significant main effect of baseline noise (normalized to EMGmax) between the lEMG and the Telemyo 900 systems (0.27 ± 0.17% EMGmax vs. 0.19 ± 0.08% EMGmax, respectively; p = .174; Fig. 4). However, there was a significant interaction between EMG amplifier and muscle (p = 0.033) with greater baseline noise in TA when detected with lEMG compared with Telemyo 900 (p = 0.045), though the magnitude of the difference was rela-
2.5. Statistical analysis Within-subjects repeated measures ANOVA was used to test for differences in EMG parameters between the lEMG and Telemyo 900 units. For all analyses, within-subjects factors included the EMG system (lEMG or Telemyo 900) with additional factors and dependent variables as described below. For EMG amplitudes during MVCs, within-subjects factors included MVC time (start/end) and muscle (BB/TA). For EMG amplitudes during submaximal force contractions, within-subjects factors included contraction intensity (25%, 50%, and 75% MVC) and muscle. For BB EMG amplitudes
Fig. 3. Absolute EMG amplitudes during MVC. There were no differences in absolute EMG amplitudes (mV) for MVCs at the start and end of the experiment (p = .088). Absolute EMG amplitudes were greater while using the Telemyo 900 compared with lEMG, likely because lEMG used more selective concentric electrodes. In addition, absolute EMG amplitudes were greater for biceps brachii (BB) compared with tibialis anterior (TA). Values are means ± SE. ap = .036 vs. lEMG; bp = .012 vs. TA.
Please cite this article in press as: Walters TJ et al. Validation of a portable EMG device to assess muscle activity during free-living situations. J Electromyogr Kinesiol (2013), http://dx.doi.org/10.1016/j.jelekin.2013.06.004
T.J. Walters et al. / Journal of Electromyography and Kinesiology xxx (2013) xxx–xxx
Fig. 4. Baseline noise from lEMG and Telemyo 900. There were no differences in baseline noise, normalized as a percent of maximal EMG, between the two EMG systems (p = .174). However, there was an EMG amplifier by muscle interaction (p = 0.033), with greater baseline noise in tibialis anterior (TA) when detected with lEMG. In addition, there was an EMG amplifier by time interaction (p = 0.003) with greater baseline noise at the start of the experiment when detected with lEMG. Nonetheless, baseline noise was always below 0.5% maximum EMG (EMGmax). Values are means ± SE. ap = 0.045 vs. baseline noise detected in TA with Telemyo 900; bp = 0.043 vs. baseline noise detected at the start of the experiment with Telemyo 900.
5
Fig. 5. EMG amplitudes during 25%, 50%, and 75% MVC contractions. There were no significant differences in EMG amplitudes between the lEMG and Telemyo 900 systems (p = .936), and all other main effects and interactions were not significant (p > .094). EMG amplitudes detected simultaneously from the two systems were plotted for both muscles. As expected there was variability in individual values, however the slope of the best-fit line was 1.00 and the R2 value was 0.82. Also note that subjects performed the task without visual feedback and frequently underestimated the amount of force necessary for the task, contributing to the large range of forces produced.
tively small at 0.15% EMGmax. In addition, we found a significant interaction between EMG amplifier and time (p = 0.003) with greater baseline noise at the start of the experiment when detected with lEMG compared with Telemyo 900 (p = 0.043), though the magnitude of the difference was also small and 0.13% EMGmax. No other interaction was significant (p > .244). Although there were significant differences between the two EMG systems in terms of baseline noise, the differences were below 0.15% EMGmax and baseline noise did not exceed 0.5% EMGmax.
3.3. Submaximal contractions There were no significant differences for the three submaximal force contractions between the lEMG and the Telemyo 900 systems (p = .936), and all other main effects and interactions were not significant (p > .094). Note that subjects performed the submaximal contractions without visual feedback, thus there was large variability in the submaximal forces actually produced, with subjects generally producing force magnitudes less than the target force level. Fig. 5 is a scatter plot that shows the normalized EMG amplitudes from both EMG systems during the three submaximal contractions for all subjects. Although there was variability in individual values, as commonly expected with EMG and performing the task without feedback, the slope of the line representing the best linear fit of the data is 1.00 (Fig. 5), and the R2 value is 0.82, suggesting that the EMG systems recorded muscle activity similarly across varying magnitudes of muscle contraction.
3.4. Simulated activities of daily living There were no significant differences for EMGs from BB during the ‘‘upper’’ body simulated ADLs between lEMG and Telemyo 900 systems (4.50 ± 2.46% EMGmax vs. 5.21 ± 3.05% EMGmax; p = .488, respectively; Fig. 6A), nor for EMGs from TA during the ‘‘lower’’ body ADLs (13.44 ± 7.12% EMGmax vs. 9.81 ± 2.45% EMGmax; p = .199, respectively; Fig. 6B). There was a significant main effect for task in TA (p = .037) but not for BB (p = .483; Fig. 6). EMG amplifier by task interactions for BB and TA were not significant
Fig. 6. EMG amplitudes during simulated activities of daily living. There were no significant differences between lEMG and Telemyo 900 systems for (A) EMG amplitudes from biceps brachii (BB) during the ‘‘upper’’ body tasks (p = .488), nor for (B) EMG amplitudes from tibialis anterior (TA) during the ‘‘lower’’ body tasks (p = .199). All interactions including the EMG amplifier were not significant (p > .127) indicating the EMG systems performed similarly across tasks.
Please cite this article in press as: Walters TJ et al. Validation of a portable EMG device to assess muscle activity during free-living situations. J Electromyogr Kinesiol (2013), http://dx.doi.org/10.1016/j.jelekin.2013.06.004
6
T.J. Walters et al. / Journal of Electromyography and Kinesiology xxx (2013) xxx–xxx
(p = .263, p = .127, respectively), which indicated that the EMG systems performed similarly during each individual task. 3.5. Free living There were no significant differences between the lEMG and the Telemyo 900 systems for normalized EMG amplitudes (p = .667) nor for EMG durations (p = .581) during the simulated free-living period (Fig. 7). There was a significant main effect for threshold in EMG amplitudes and durations during free living (p < .001, p = .001, respectively). All interactions for normalized EMG amplitudes and EMG durations were not significant (p > .050). As the use of Telemyo 900 restricted recording EMGs within 300 yards of the laboratory, we tested the feasibility of using lEMG for a longer wear time during a free-living period outside the laboratory. Fig. 8 depicts representative EMG data recorded from BB and TA and accelerations from a hip-worn accelerometer (GT3X+; Actigraph, Pensacola, FL). Although a comparison to Telemyo 900 was not possible, lEMG successfully recorded EMGs during the longer wear time. 4. Discussion This is the first study to systematically validate a low-cost EMG data logging system designed specifically for free-living situations.
Fig. 8. Representative data during a longer free-living period performed outside of the laboratory. To test the feasibility of recording EMGs for longer periods of time with lEMG, EMGs were detected from (A) biceps brachii and (B) tibialis anterior during a 7 h free-living period. In addition, average counts/min from a (C) hipworn accelerometer was also computed. EMG and accelerometry data was collected simultaneously and is shown in 1-s epochs.
Results from the current study indicate that the portable, low-cost lEMG and the laboratory-based Telemyo 900 system performed similarly in the current study, supporting the use of lEMG for recording EMGs during free living situations. Strengths and limitations of lEMG are discussed below. 4.1. Strengths of lEMG First, similarities across the two EMG systems in the current study in terms of EMG parameters during submaximal contractions, simulated ADLs, and free-living support the use of lEMG. As clearly represented in Fig. 1, the EMGs from the two systems are nearly identical and there was no clear artifact while recording from either system. Second, similar to using accelerometers outside of the laboratory to track physical activity (Swartz et al., 2000; Troiano et al., 2008), the low-cost and small size of lEMG allows deployment of multiple units simultaneously from the laboratory or a remote location. In addition, lEMG can be deployed without a large financial risk as is common when sending higher-cost, laboratory-based EMG systems into free-living situations. The ease of use and deployment of lEMG should expand the application of long-term EMG within the fields of aging, ergonomics, and clinical disorders. Third, the design of the concentric electrode for the lEMG system enhances its practicality in free-living situations. Advantages of the concentric electrode over traditional bipolar recordings include: easy application of the electrode as its orientation with respect to muscle fiber angle or changes in fiber angle will have minimal influence on the EMG signal (Farina and Cescon, 2001; Rainoldi et al., 2004); higher spatial selectivity with respect to traditional detection systems (Farina and Cescon, 2001) and thereby reducing the potential confound of crosstalk (De Luca and Merletti, 1988; Farina et al., 2002; Solomonow et al., 1994); and the concentric surface electrode used with lEMG has a large adhesive surface with a short cable length, which reduces the problem of electrode adherence during long-term recordings and potential artifacts due to cable movement.
Fig. 7. Free-living EMG amplitudes and durations. EMG amplitudes and durations were estimated using three different analysis thresholds (2%, 3%, and 4% EMGmax; see Section 2). There were no significant differences between the two EMG systems for (A) EMG amplitudes (p = .667), nor for (B) EMG durations (p = .581). All interactions for EMG amplitudes and EMG durations were not significant (p > .050, p > .073, respectively).
4.2. Limitations of lEMG First, baseline noise was not identical across both EMG systems. Specifically, there was greater baseline noise in TA when detected
Please cite this article in press as: Walters TJ et al. Validation of a portable EMG device to assess muscle activity during free-living situations. J Electromyogr Kinesiol (2013), http://dx.doi.org/10.1016/j.jelekin.2013.06.004
T.J. Walters et al. / Journal of Electromyography and Kinesiology xxx (2013) xxx–xxx
with lEMG compared with Telemyo 900 (p = 0.045) and greater baseline noise at the start of the experiment when detected with lEMG compared with Telemyo 900 (p = 0.043; Fig. 4). It is not clear why these differences exist; however, the magnitude of the differences across systems was small (<0.15% EMGmax) and baseline noise for both systems was less than 0.5% EMGmax. Nonetheless, although amplitude thresholds at 2% EMGmax are most commonly used in long-term EMG studies (Kern et al., 2001; Klein et al., 2010; Theou et al., 2010), some studies use lower thresholds at 0.5% EMGmax to investigate how sustained low-activity muscle activation relates to pain (Mork and Westgaard, 2005; Ostensvik et al., 2009). Our study suggests caution in deploying lEMG with such low-amplitude thresholds, at least for some muscles like TA where baseline noise can approach 0.5% EMGmax (Fig. 4). Second, unlike all laboratory-based EMG systems, instantaneous visual feedback of EMG is not possible with lEMG. Thus, to visually inspect EMG signals from the lEMG, the experimenter must eject the microSD card and load it onto a computer with the OT Biolab software. Third, since lEMG exhibits a limited output range compared to laboratory-based systems (±3.3 V vs. ±5 V, respectively), and there is no adjustable gain with lEMG, saturation of the EMG signal is more likely. For example, in early piloting of lEMG we tested vastus medialis but found the signal saturated during some MVC trials. However, signal saturation was not an issue for BB or TA. This issue could limit the usefulness of lEMG when recording EMGs from other muscles. Nonetheless, normalization of EMG amplitude to submaximal contractions is likely a viable solution (Yang and Winter, 1983) as free-living muscle activity is commonly performed at low intensity levels (Kern et al., 2001; Klein et al., 2010). In summary, this was the first study to validate a low-cost EMG data logging system specifically designed for use outside of the laboratory. Comparison between a low-cost system and the laboratory-based system revealed similarities in several outcome variables, which suggests that the lEMG system is a valid approach to recording EMGs during free-living situations. Though the lEMG system exhibits certain constraints compared to the Telemyo 900 system, specifically related to baseline noise, the design of the lEMG system makes it a viable approach for large-scale investigation of long-term, free-living neuromuscular function across different subject populations (e.g., healthy, elderly, and stroke). Similar to the miniaturization and common deployment of accelerometers to measure physical activity, low-cost, portable EMG systems specifically designed for free-living situations will allow researchers to better assess the functional use of muscles during activities of daily living.
7
Kern DS, Semmler JG, Enoka RM. Long-term activity in upper- and lower-limb muscles of humans. J Appl Physiol 2001;91:2224–32. Klein CS, Peterson LB, Ferrell S, Thomas CK. Sensitivity of 24-h EMG duration and intensity in the human vastus lateralis muscle to threshold changes. J Appl Physiol 2010;108:655–61. Monster AW, Chan H, O’Connor D. Activity patterns of human skeletal muscles: relation to muscle fiber type composition. Science 1978;200:314–7. Mork PJ, Westgaard RH. Long-term electromyographic activity in upper trapezius and low back muscles of women with moderate physical activity. J Appl Physiol 2005;99:570–8. Ostensvik T, Veiersted KB, Nilsen P. A method to quantify frequency and duration of sustained low-level muscle activity as a risk factor for musculoskeletal discomfort. J Electromyogr Kinesiol 2009;19:283–94. Rainoldi A, Melchiorri G, Caruso I. A method for positioning electrodes during surface EMG recordings in lower limb muscles. J Neurosci Methods. 2004;134:37–43. Saitou K, Masuda T, Michikami D, Kojima R, Okada M. Innervation zones of the upper and lower limb muscles estimated by using multichannel surface EMG. J Hum Ergol (Tokyo) 2000;29:35–52. Solomonow M, Baratta R, Zhou B, Lu Y, Zhu M, Acierno S. Surface and wire EMG crosstalk in neighbouring muscles. J Electromyogr Kinesiol 1994;4:131–42. Swartz AM, Strath SJ, Bassett Jr DR, O’Brien WL, King GA, Ainsworth BE. Estimation of energy expenditure using CSA accelerometers at hip and wrist sites. Med Sci Sports Exerc 2000;32:S450–6. Theou O, Jones GR, Vandervoort AA, Jakobi JM. Daily muscle activity and quiescence in non-frail, pre-frail, and frail older women. Exp Gerontol 2010;45:909–17. Troiano RP, Berrigan D, Dodd KW, Masse LC, Tilert T, McDowell M. Physical activity in the United States measured by accelerometer. Med Sci Sports Exerc 2008;40:181–8. Yang JF, Winter DA. Electromyography reliability in maximal and submaximal isometric contractions. Archiv Phys Med Rehab 1983;64:417–20.
Tygh J. Walters received his B.A. in Exercise Physiology and Human Performance from Ripon College in Ripon, WI in 2009 prior to completing his M.S. degree in Kinesiology from the University of WisconsinMilwaukee in 2012. His research interests focus on physical function and long-term electromyography in free-living situations. He is currently an assistant at an interventional pain clinic in Norfolk, NE.
Kayla Kaschinske received her B.S. in Kinesiology at the University of Wisconsin – Milwaukee in 2011. She is currently a Doctor of Physical Therapy student at the University of Wisconsin–Madison and expects to graduate in May of 2014.
Acknowledgement UW-Milwaukee Research Growth Initiative supported the study but had no other involvement. References De Luca CJ, Gilmore LD, Kuznetsov M, Roy SH. Filtering the surface EMG signal: movement artifact and baseline noise contamination. J Biomech 2010;43:1573–9. De Luca CJ, Merletti R. Surface myoelectric signal cross-talk among muscles of the leg. Electroencephalogr Clin Neurophysiol 1988;69:568–75. Farina D, Cescon C. Concentric-ring electrode systems for noninvasive detection of single motor unit activity. IEEE Trans Biomed Eng 2001;48:1326–34. Farina D, Merletti R, Indino B, Nazzaro M, Pozzo M. Surface EMG crosstalk between knee extensor muscles: experimental and model results. Muscle Nerve 2002;26:681–95. Jakobi JM, Edwards DL, Connelly DM. Utility of portable electromyography for quantifying muscle activity during daily use. Gerontology 2008;54:324–31. Keenan KG, Collins JD, Massey WV, Walters TJ, Gruszka HD. Coherence between surface electromyograms is influenced by electrode placement in hand muscles. J Neurosci Methods 2011;195:10–4.
Scott J. Strath received his B.Ed. in Physical Education (1996) from Sheffield Hallam University, his M.S. in Exercise Science (1998) from Ball State University, and his Ph.D. in Exercise Science from the University of Tennessee (2001). He then completed post-doctoral training in the Department of Physical Medicine and Rehabilitation at the University of Kentucky and the University of Michigan. He is currently an Associate Professor in the Department of Kinesiology at the University of Wisconsin-Milwaukee. His research interests revolve around physical activity and public health. Specific interests include the relationship between physical activity and cardiovascular health; physical activity promotional strategies; physical activity assessment; physical activity epidemiology; community and national physical activity patterns; and environmental determinants of physical activity behavior.
Please cite this article in press as: Walters TJ et al. Validation of a portable EMG device to assess muscle activity during free-living situations. J Electromyogr Kinesiol (2013), http://dx.doi.org/10.1016/j.jelekin.2013.06.004
8
T.J. Walters et al. / Journal of Electromyography and Kinesiology xxx (2013) xxx–xxx
Ann M. Swartz received her B.S. in Kinesiology (1995) from the University of Wisconsin-Madison, and M.S. in Exercise Science (1998) from Ball State University, and her Ph.D. in Exercise Science (2001) from the University of Tennessee. She then completed post-doctoral training in the Department of Physical Medicine and Rehabilitation at the University of Kentucky and the University of Michigan. She is currently an Associate Professor in the Department of Kinesiology at the University of Wisconsin-Milwaukee. Her research interests center on obesity; in particular, the health benefits of physical activity for overweight and obese individuals including 1) relationships between physical activity, health and obesity level, 2) novel translational physical activity interventions to improve health for overweight and obese individuals, and 3) the accurate assessment of physical activity behavior.
Kevin G. Keenan received his B.S. (1990) and M.S. (1998) degrees in Kinesiology from the University of Michigan. He subsequently worked as the Head Strength and Conditioning Coach at Providence College prior to receiving his Ph.D. in Integrative Physiology from the University of Colorado-Boulder (2005) and completing post-doctoral training (2005–2008) at Cornell University in the Sibley School of Mechanical and Aerospace Engineering. He is currently an Assistant Professor in the Department of Kinesiology at the University of Wisconsin-Milwaukee. His research interests focus on understanding the interplay between motor function, physical activity, and health using high-density surface EMG arrays, long-term (>6 h) surface EMG recordings, indwelling EMG recordings, and computational modeling.
Please cite this article in press as: Walters TJ et al. Validation of a portable EMG device to assess muscle activity during free-living situations. J Electromyogr Kinesiol (2013), http://dx.doi.org/10.1016/j.jelekin.2013.06.004