- Email: [email protected]
Surface-EMG analysis for the quantification of thigh muscle dynamic co-contractions during normal gait
Gait & Posture 51 (2017) 228–233
Contents lists available at ScienceDirect
Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost
Full length article
Surface-EMG analysis for the quantification of thigh muscle dynamic co-contractions during normal gait Annachiara Strazzaa , Alessandro Mengarellia , Sandro Fiorettia , Laura Burattinia , Valentina Agostinib , Marco Knaflitzb , Francesco Di Nardoa,* a b
Department of Information Engineering, Università Politecnica delle Marche, 60131 Ancona, Italy Department of Electronics and Telecommunications, Politecnico di Torino, Torino, Italy
A R T I C L E I N F O
Article history: Received 27 May 2016 Received in revised form 26 October 2016 Accepted 1 November 2016 Keywords: sEMG Knee Statistical gait analysis Antagonist muscles Co-contraction
A B S T R A C T
The research purpose was to quantify the co-contraction patterns of quadriceps femoris (QF) vs. hamstring muscles during free walking, in terms of onset-offset muscular activation, excitation intensity, and occurrence frequency. Statistical gait analysis was performed on surface-EMG signals from vastus lateralis (VL), rectus femoris (RF), and medial hamstrings (MH), in 16315 strides walked by 30 healthy young adults. Results showed full superimpositions of MH with both VL and RF activity from terminal swing, 80 to 100% of gait cycle (GC), to the successive loading response (0–15% of GC), in around 90% of the considered strides. A further superimposition was detected during the push-off phase both between VL and MH activation intervals (38.6 12.8% to 44.1 9.6% of GC) in 21.9 13.6% of strides, and between RF and MH activation intervals (45.9 5.3% to 50.7 9.7 of GC) in 32.7 15.1% of strides. These findings led to identify three different co-contractions among QF and hamstring muscles during able-bodied walking: in early stance (in 90% of strides), in push-off (in 25–30% of strides) and in terminal swing (in 90% of strides). The co-contraction in terminal swing is the one with the highest levels of muscle excitation intensity. To our knowledge, this analysis represents the first attempt for quantification of QF/hamstring muscles co-contraction in young healthy subjects during normal gait, able to include the physiological variability of the phenomenon. ã 2016 Elsevier B.V. All rights reserved.
1. Introduction Position and complexity of knee joint suggest that it is relatively weak and susceptible to injury [1]. Thus, knee joint relies on muscles and ligaments to ensure stability. Main muscles carrying out this task are quadriceps femoris (QF) and hamstrings muscles. QF is the most direct source of extensor control [2]. During walking, contraction of quadriceps muscles represents the primary absorbing mechanism of impact during weight acceptance [2]. In stance, extensors act to decelerate knee flexion, while in swing they contribute to limb progression [2]. At knee joint, hamstrings are mainly involved in knee flexion control. During walking, hamstrings serve as a useful protective function in preventing knee
Abbreviations: F, flat foot contact; GC, gait cycle; H, heel contact; MH, medial hamstrings; P, push-off; QF, quadriceps femoris; RF, rectus femoris; S, swing; SGA, statistical gait analysis; VL, vastus lateralis. * Corresponding author at: Department of Information Engineering, Università Politecnica delle Marche, Via Brecce Bianche, 60131, Ancona, Italy. E-mail address: [email protected] (F. Di Nardo). http://dx.doi.org/10.1016/j.gaitpost.2016.11.003 0966-6362/ã 2016 Elsevier B.V. All rights reserved.
hyperextension in terminal swing and in following early stance [2]. QF and hamstring are antagonist muscle groups for knee joint during walking. Knee-joint stability is mainly preserved by co-contractions of antagonist muscle groups [3]. Co-contraction generally-held purpose is to augment ligament function in maintenance of joint stability, providing resistance to rotation at a joint and equalizing pressure distribution at joint surfaces [3]. Quantitative estimation of myoelectric activity of QF and hamstring muscles and of their concomitant recruitment could be very useful for analyzing kneejoint stability. Different studies assessed QF and hamstring muscles co-contractions during able-bodied walking, starting from the activation picture of rectus femoris, vastii and hamstring muscles [1,4–6]. Recent studies based on statistical gait analysis (SGA) of numerous strides/subject, reported that a single muscle showed a different number of activation intervals in different strides of the same walking [7–11]. Five different activation modalities were identified for both rectus femoris [8] and vastii [7], describing the large variability of muscular recruitment. This led to identifying muscular activities which are usually not reported in classic
A. Strazza et al. / Gait & Posture 51 (2017) 228–233
reference studies for EMG during gait [2,12], as for example VL activity in push-off phase [7,9]. Moreover, SGA-based studies also reported the occurrence frequency, defined as the frequency each muscle-activation occurs, quantified by the number of strides in which muscle is recruited with that specific activation modality [10]. This is a parameter hardly ever considered, because of low number of strides analyzed in traditional EMG studies. From this point of view, data on QF and hamstring muscles co-contractions reported by cited studies [1,4–6] are limited by the small number of gait cycles (GC) considered during an assessment session. The aim of the study was the quantitative assessment during able-bodied walking of the concomitant recruitment (i.e. cocontraction) of QF and hamstrings muscles in numerous (hundreds) strides per subject, in terms of variability of onset-offset muscular activation, excitation intensity, and occurrence frequency. Vastus lateralis (VL) and rectus femoris (RF) were analyzed as representative muscles for QF group. Medial hamstrings (MH) were considered as representative of hamstrings group. The technique of SGA [13,14] was performed to align with recent literature in handling numerous strides. 2. Materials and methods 2.1. Subjects Thirty healthy adults (15 females and 15 males) were recruited. Mean (SD) characteristics are: age = 23.8 1.9 years; height = 173 10 cm; weight = 63.3 12.4 kg; body mass index (BMI) = 20.8 2.1 kg m 2. Exclusion criteria included joints pain, neurological pathology, orthopedic surgery, abnormal gait or BMI 25. Participants signed informed consent. The research was undertaken in compliance with ethical principles of Helsinki Declaration and approved by institutional expert committee. 2.2. Signal acquisition and processing Signals were acquired (sampling rate:2 kHz; resolution:12 bit) and processed by the multichannel recording system, Step32 (Version PCI-32 ch2.0.1. DV, MedicalTechnology, Italy). Each subject was instrumented with foot-switches, knee electrogoniometers and sEMG electrodes on both lower limbs. Three foot-switches (size:11 11 0.5 mm; activation force:3 N) were attached beneath heel, first and fifth metatarsal heads of each foot. An electro-goniometer (accuracy:0.5 ) was attached to lateral side of each lower limb for measuring knee-joint angles in sagittal plane.; sEMG signals were detected with single-differential electrodes with fixed geometry (Ag/Ag-Cl disks, size:7 27 19 mm, electrode diameter:4 mm, interelectrode distance:12 mm,
229
gain:1000, high pass filter:10 Hz, input impedance >1.5 GV, CMRR> 126 dB, input referred noise 1 mVrms), and with variable geometry (Ag/Ag-Cl disks, minimum interelectrode distance:12 mm, gain:1000, high-pass filter:10 Hz, input impedance >1.5 GV, CMRR >126 dB, input referred noise 200 nVrms). sEMG signals were further amplified and low-pass filtered (450 Hz) by recording system.; Skin was shaved, cleansed with abrasive paste and wet with a soaked cloth. Electrodes were applied over VL, MH and RF, following SENIAM recommendations for electrode locationorientation over muscles with respect to tendons, motor points and fiber direction [15]. Participant set-up is shown in Fig. 1. Subjects were asked to walk barefoot overground for 5 min at natural pace, back and forth over a 12-meters-straight track, maintaining a steady speed. Natural pace was chosen because walking at comfortable speeds improves repeatability of EMG data [16]. Footswitch signals were debounced, converted to four levels, Heel contact (H), Flat foot contact (F), Push-off (P), Swing (S), and processed to segment and classify different GCs [17].; Electrogoniometric signals were low-pass filtered (FIR filter, 100 taps, cutoff frequency:15 Hz). Knee angles along with sequences and durations of gait phases derived by basographic signal were used by a multivariate statistical filter, to detect and discard outlier cycles like those relative to deceleration, reversing, and acceleration. sEMG signals were high-pass filtered (FIR filter, 100 taps, cutoff frequency:20 Hz) and processed by a double-threshold statistical detector, allowing a user-independent assessment of activation intervals [18]. This technique consists of selecting a first threshold z and observing m successive samples: if at least r0 (second threshold) out of successive m samples are above z, presence of the signal is acknowledged. Details are reported in [18]. Muscular co-contractions were quantified by assessing the overlapping period among activation intervals of considered muscles, in the very same strides [19]. Overlapping periods 30 ms were not considered in co-contraction computation, since a muscle activation 30 ms has no effect in controlling joint motion during gait [18]. 2.3. Statistical gait analysis Statistical gait analysis (SGA) performs a statistical characterization of gait analyzing spatial-temporal and sEMG-based parameters over numerous strides, during the same walking trial [7–11,13,14,20–24]. SGA is based on the observation that the number of muscle activations within a cycle is cycle dependent: it may vary from stride to stride. Therefore, averaging across strides should be performed only over those strides sharing the same number of activations, i.e. belonging to the same activation modality. Activation modality is defined as the number of times a
Fig. 1. Participant set-up.
230
A. Strazza et al. / Gait & Posture 51 (2017) 228–233
muscle activates during a single GC (n-activation modality consists of n activation intervals for the considered muscle, during a single GC). Mean activation intervals (normalized with respect to GC) for each activation modality are achieved, according to following steps. First, muscle activation intervals relative to each GC are identified, computing muscle onset/offset instants in temporal space [18], as previously described. Then, for each subject, gait cycles are grouped according to their activation modality, and onset-offset timings are averaged separately for each modality. Eventually, onset/offset time instants of each activation modality are averaged over the thirty subjects. Averaged onset/offset percentage time instants are normalized with respect to GC to provide mean activation intervals in percentage of GC. SGA was performed by Step32 system. SGA analysis of numerous strides allowed quantifying thigh co-contractions in terms of onset-offset muscular activation, quantized excitation intensity, and occurrence frequency (i.e. the number of times muscle is recruited with a specific activation modality during a single walking trial). Further details can be found in Supplementary material. 2.4. Assessment of excitation intensity A statistical representation of mean muscle excitation intensity over population was achieved as follows. Step32 provides the excitation intensity for each subject, quantized in three levels. In this study, the three intensity levels, considered in ascending order, were associated with the value of 1, 2 and 3, respectively. A global quantized mean signal over all subjects, ranging from 0 (no signal) to 90 (maximum value), was obtained for each percentage unit of GC, as the sum of intensity levels of every single mean signal. This global mean signal was subdivided in four intervals ranging from 0 to 12 (no muscle activation), from 13 to 30 (low-level activation), from 31 to 60 (medium-level activation), and from 61 to 90 (highlevel activation). Mean activation intervals are then represented by horizontal bars using a three-level grey scale for the excitation intensity; darker shadowing corresponds to increasing intensity of the mean signal, i.e. low-level (light grey), medium-level (dark grey), and high level (black) activation. The purpose of this quantization is mainly to identify the peak regions of myoelectric activity. 2.5. Statistics Data are reported as means standard deviation (SD). Lilliefors test was used to evaluate if each data vector had a normal distribution. The analysis of variance (ANOVA) test, followed by a multiple-comparison test, was used to compare normally distributed samples. To determine the power of an observed effect based on the sample size and parameter estimates derived from data set, the statistical power analysis was performed. Statistical significance was set at 5%. 3. Results Mean of 454 112 strides was considered for each subject. Durations of gait phases ware computed; H-phase lasts for 5.6 1.7% of GC, F-phase 28.9 6.0%, P-phase 23.4 4.8%, and Sphase 44.1 2.8%. Activation modalities for each muscle are reported in Fig. 2. Mean results are reported with data from right and left lower limb considered all together.; Different superimposition intervals between VL and RF activities were detected during GC. In strides characterized by RF double activation (Fig. 3A), superimpositions between RF and VL activation intervals were observed in early stance (1.9 5.4 to 23.3 14.1% of GC) and in terminal swing (81.4 2.9% to GC end). These superimpositions were detected in the two main activation modalities for VL. Also in
Fig. 2. Mean activation intervals vs. percentage of gait cycle for modalities with 1, 2, 3, 4 and 5 activations detected during walking for VL (panel A), RF (panel B), and MH (panel C). H, F, P and S phases are delimited by dashed vertical lines.
strides where RF adopted the 3- and 4-activation modalities, superimpositions were observed from terminal swing to the following early stance, for all VL activation modalities (from the beginning to 15.9 5.0% and from 85.6 3.3 to 99.6 0.8% of GC for RF triple activation, Fig. 3B; for RF 4-activation from the beginning to 14.3 3.3% and from 85.6 3.2% to the end of GC, Fig. 3C). In strides characterized by RF 4-activation (Fig. 2C), a further superimposition between VL and RF activation intervals was observed during mid-stance from 36.2 6.8% to 39.3 4.4% of GC, in 6.9 4.9% of considered strides. Occurrence frequencies of RF/VL co-contractions are reported in Fig. 3D. Different superimposition intervals were detected also between VL and MH activity. In strides with VL double activation (Fig. 4A), superimpositions with the two main MH activation intervals were observed from the beginning to 14.5 10.2% of GC) and from 81.5 3.3% to GC end. Also in strides where VL adopted 3(Fig. 4B) and 4-activation modality (Fig. 4C), VL/MH superimpositions were observed from terminal swing to the following early stance, for all MH activation modalities (from the beginning to 15.5 15.1% and from 81.8 3.8 to 99.9 0.1% of GC for VL triple activation; for VL 4-activation from the beginning to 14.8 14.2%
A. Strazza et al. / Gait & Posture 51 (2017) 228–233
231
and 4-activation (from the beginning of GC to 14.3 3.3%) modalities of MH and during swing, for 2-activation (from 80.6 10.7% to 98.2 4.8%), 3-activation (from 85.6 3.3% to the end of GC) and 4-activation (from 85.6 3.2% to 97.8 4.2%) modalities of MH. When RF increases to 3-activation modality, a co-contraction with MH occurs from 45.9 5.3% to 50.7 9.7 of GC (Fig. 5B). Occurrence frequencies of RF/MH co-contractions are reported in Fig. 5D. Further information on muscle co-contraction was given in terms of excitation intensity. Mean values over 30 subjects of muscle excitation intensity were computed following the procedure reported in “Assessment of excitation intensity” section and depicted by means of a three-level grey scale in Figs. 3–5. Percentage occurrence frequencies of co-contractions were compared. No significant difference was detected among occurrence frequencies of RF/VL (100%), VL/MH (90.3 13.3%), and RF/ MH (92.4 16.6%) co-contractions in early stance and in swing. RF/ MH co-contraction in push-off showed a significant higher occurrence frequency respect to the other push-off co-contractions. Results of statistical power analysis showed that number of participants (30) guarantees a power 0.90 for every difference tested in the study. 4. Discussion
Fig. 3. Mean values of VL activation intervals vs. percentage of gait cycle, detected in strides where RF shows 2-activation modality (panel A), 3-activation modality (panel B), and 4-activation modality (panel C). VL activation intervals are reported separately for modalities with 2 and 3 activations. Mean excitation intensity are represented by using a three-level grey scale; darker shadowing corresponds to increasing intensity of the mean signal, i.e. low-level (light grey), medium-level (dark grey), and high-level (black) activation. VL/RF co-contractions are highlighted by box. Mean (+SD) percentage occurrence frequency of each of the three different VL/RF co-contractions are reported in panel D. H, F, P and S phases are delimited by dashed grey vertical lines.
and from 81.5 4.1% to the end of GC). In strides with VL triple activation (Fig. 4B), a further VL/MH superimposition was observed from 38.6 12.8% to 44.1 9.6% of GC. Occurrence frequencies of VL/MH co-contractions are reported in Fig. 4D. Superimpositions between sEMG activity of RF and MH were also quantified. In strides where RF presented a double, triple, and quadruple activation (Fig. 5), a RF/MH superimposition was observed during early stance, for 2-activation (from 1.9 5.4% to 12.5 8.6%), 3-activation (from the beginning of GC to 12.8 6.3%)
In agreement with [7,8,11,20], this study showed that muscles adopt different activation modalities in different strides of the same walking. This finding was reported also for ankle muscles [21–23] and in children [9,24]. This means that a muscle could be recruited in a single stride with one, two, three or even more activities and in the following stride the number of activities could be different. Numerous strides allowed identifying five different activation modalities for each muscle, describing the large variability of muscular recruitment (Fig. 2). This leads to identifying muscular activities usually not reported in classic reference studies for EMG during gait [2,12], as VL activity in pushoff phase. The study of the whole variability of the three muscles allowed evaluating superimposition of muscle activity in all activation modalities, providing what the present study suggests to be the complete picture of co-contractions among VL, RF and MH. Optimal number of strides for providing a suitable average EMG profile was suggested to be 20, for natural walking speed [25]. Since averaged values were computed separately over five activation modalities for each muscle (Fig. 2), a total number of at least 100 strides per subject should be analyzed for a complete characterization of muscle recruitment variability. This matches also with our practical experience. RF-activity from terminal swing to the following loading response (Fig. 2) was detected in the nearly totally of strides. Main VL-activity occurred at the same relative percentage in GC as RF (Fig. 2), determining a superimposition in 100% of strides. These activities are recognized as typical activations of QF muscles as knee extensors. The superimposition during early stance could be intended as a synergic action of muscles for controlling weight bearing; this matches with [9]. Similarly, the synergic action of VL and RF in terminal swing is suggested to be present to assist knee extension and develop muscle tension for weight acceptance during loading response. However RF-signal, detected from terminal swing to the following loading response, could also be due to cross-talk from vastii [8,26]. High levels of muscle excitation intensity (Fig. 3) in these gait phases could support both hypotheses. MH was also recruited (Fig. 2) in the same percentage of GC where the above-cited activity of QF muscles was detected. This determined a superimposition of MH with both VL and RF activity in more than 90% of considered strides (Figs. 4–5). While it is
232
A. Strazza et al. / Gait & Posture 51 (2017) 228–233
Fig. 4. Mean values of MH activation intervals (black bars) vs. percentage of gait cycle, detected in strides where VL (grey bars) shows 2-activation modality (panel A), 3-activation modality (panel B), and 4-activation modality (panel C). MH activation intervals are reported separately for modalities with 2 and 3activations. Mean excitation intensity are represented by using a three-level grey scale; darker shadowing corresponds to increasing intensity of the mean signal, i.e. low-level (light grey), medium-level (dark grey), and high-level (black) activation. VL/MH cocontractions are highlighted by box Mean (+SD) percentage occurrence frequency of each of the three different VL/MH co-contractions are reported in panel D. H, F, P and S phases are delimited by dashed grey vertical lines.
reasonable indicating that mono-articular vastii muscles influence actions about the knee, it is real hard isolating the influence of a biarticular muscle (MH) to a single joint. Thus, interpretation of superimposition in early stance and swing needs to be deepened. However, superimposition in terminal swing could be reasonably
Fig. 5. Mean values of MH activation intervals (black bars) vs. percentage of gait cycle, detected in strides where RF (grey bars) shows 2-activation modality (panel A), 3-activation modality (panel B), and 4-activation modality (panel C). MH activation intervals are reported separately for modalities with 2 and 3 activations. Mean excitation intensity are represented by using a three-level grey scale; darker shadowing corresponds to increasing intensity of the mean signal, i.e. low-level (light grey), medium-level (dark grey), and high-level (black) activation. VL/MH cocontractions are highlighted by box. Mean (+SD) percentage occurrence frequency of each of the three different RF/MH co-contractions are reported in panel D. H, F, P and S phases are delimited by dashed grey vertical lines.
intended as an action of muscles across the same joint, the knee. It likely occurs in this gait phase to assist knee extension, developing muscle tension for weight acceptance during loading response [2,3]. This is the co-contraction with the highest level of muscle excitation intensity (Figs. 4–5). Moreover, it is involved in assisting the anterior cruciate ligament (ACL) to prevent excessive anterior
A. Strazza et al. / Gait & Posture 51 (2017) 228–233
tibial displacement and in protecting ACL itself from injury [1]. This supports the hypothesis of a regulatory role of the co-contraction between QF and hamstring muscles in providing knee joint stability [3]. RF and VL overlapped their activity also in push-off phase (Fig. 3C). RF and VL are likely working in synergy for modulating rapid knee flexion [7–9,27,28], although in only 8.14 15.7% of considered strides and with low levels of muscle excitation intensity (Fig. 3C). At the same time, VL is able to contribute to patella stabilization before entering pre-swing phase and RF participates to hip flexion [7–9,27,28]. Different tasks of muscles could probably explain the large variability of muscle activity and co-contraction in this phase. During push-off phase, uncommon and variable activities of VL and MH were detected (Fig. 2). VL is likely recruited to modulate rapid knee flexion and to stabilize the patella before entering preswing phase [29]. MH acts as hip extensor to propel body forward [2]. These activities implied the appearance of a superimposition between VL and MH (Fig. 4B). It was detected in a low percentage of strides (21.9 13.6%) and with low levels of muscle excitation intensity (Fig. 4B), and showed a large variability both intra- and inter-subjects, covering the entire push-off phase (around 30–50% of GC). This superimposition should not be considered as a real cocontraction, because VL and MH are working mainly on different joints. Differently, the superimposition between sEMG signals from RF and MH (Fig. 5B), detected in 32.7 15.1% of strides, could be intended as an actual co-contraction for the control of rapid knee extension and it could play a role also in stabilization of pelvis during body progression. The levels of excitation intensity appeared to be higher than those reported for VL/MH co-activation in the same phase (Fig. 5B vs. Fig. 4B). However, outcomes on this co-contraction should be confirmed and supported by further analyses that consider a direct study of muscle function via electrical stimulation. Differences in preferred walking speed across participants could introduce variability in muscle onset-offset times. Thus, walking speed could be considered a potential confound and this could be acknowledged as a limitation of the study. To our knowledge, this analysis represents the first attempt for quantification of QF/hamstring muscles co-contraction in young healthy subjects during normal gait. Three different co-contractions were identified: in early stance (in 90% of strides), in pushoff (in 25–30% of strides) and during swing (in 90% of strides). The contribution of the study consists in providing novel quantitative information on variability of thigh-muscle co-contractions, in terms of onset-offset muscular activation, excitation intensity and occurrence frequency. Thus, the present findings can be useful in clinical context and for designing future gait studies. Conflict of interest None. Acknowledgments The authors thank Dr. Giacomo G. Ghetti (Posture and Movement Analysis Laboratory, Italian National Institute of Health and Science on Aging (INRCA), Ancona, Italy) for the precious help he gave in clinical interpretation of data. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j. gaitpost.2016.11.003.
233
References [1] S.T. da Fonseca, P.L. Silva, J.M. Ocarino, R.B. Guimaraes, M.T. Oliveira, C.A. Lage, Analyses of dynamic co-contraction level in individuals with anterior cruciate ligament injury, J. Electromyogr. Kinesiol. 14 (2004) 239–247. [2] J. Perry, Gait Analysis–Normal and Pathological Function, Slack Incorporated, Thorofare (NJ), 1992. [3] R. Baratta, M. Solomonow, B.H. Zhou, D. Letson, R. Chuinard, R. D’Ambrosia, Muscular coactivation: the role of the antagonist musculature in maintaining knee stability, Am. J. Sports Med. 16 (1988) 113–122. [4] G. Frost, J. Dowling, K. Dyson, O. Bar-Or, Cocontraction in three age groups of children during treadmill locomotion, J. Electromyogr. Kinesiol. 7 (1997) 179–186. [5] T. Hortobagyi, S. Solnik, A. Gruber, P. Rider, K. Steinweg, J. Helseth, P. DeVita, Interaction between age and gait velocity in the amplitude and timing of antagonist muscle coactivation, Gait Posture 29 (2009) 558–564. [6] A. Schmitz, A. Silder, B. Heiderscheit, J. Mahoney, Differences in lower extremity muscular activation during walking between healthy older and young adults, J. Electromyogr. Kinesiol. 19 (2009) 1085–1091. [7] F. Di Nardo, E. Maranesi, A. Mengarelli, G. Ghetti, L. Burattini, S. Fioretti, Assessment of the variability of vastii myoelectric activity in young healthy females during walking: a statistical gait analysis, J. Electromyogr. Kinesiol. 25 (2015) 800–807. [8] F. Di Nardo, S. Fioretti, Statistical analysis of surface electromyographic signal for the assessment of rectus femoris modalities of activation during gait, J. Electromyogr. Kinesiol. 23 (2013) 56–61. [9] V. Agostini, A. Nascimbeni, A. Gaffuri, P. Imazio, M.G. Benedetti, M. Knaflitz, Normative emg activation patterns of school-age children during gait, Gait Posture 32 (2010) 285–289. [10] F. Di Nardo, V. Agostini, M. Knaflitz, A. Mengarelli, E. Maranesi, L. Burattini, S. Fioretti, The occurrence frequency: a suitable parameter for the evaluation of the myoelectric activity during walking, Conf. Proc. IEEE Eng. Med. Biol. Soc. (2015) 6070–6073. [11] F. Di Nardo, A. Mengarelli, E. Maranesi, L. Burattini, S. Fioretti, Gender differences in the myoelectric activity of lower limb muscles in young healthy subjects during walking, Biomed. Signal Process. Control 19 (2015) 14–22. [12] D.H. Sutherland, The evolution of clinical gait analysis part l: kinesiological EMG, Gait Posture 14 (2001) 61–70. [13] V. Agostini, M. Knaflitz, Statistical gait analysis, in: R.U. Acharya, F. Molinari, T. Tamura, D.S. Naidu, J.S. Suri (Eds.), Distributed Diagnosis and Home Healthcare, American Scientific Publishers, Stevenson Ranch, 2012. [14] V. Agostini, F. Lo Fermo, G. Massazza, M. Knaflitz, Does texting while walking really affect gait in young adults? J. Neuroeng. Rehabil. 12 (2015) 86. [15] B. Freriks, H.J. Hermens, C. Disselhorst-Klug, G. Rau, Development of recommendations for semg sensors and sensor placement procedures, J. Electromyogr. Kinesiol. 10 (2000) 361–374. [16] M.P. Kadaba, H.K. Ramakrishnan, M.E. Wootten, J. Gainey, G. Gorton, G.V.B. Cochran, Repeatability of kinematic, kinetic, and electromyographic data in normal adult gait, J. Orthop. Res. 6 (1989) 849–860. [17] V. Agostini, G. Balestra, M. Knaflitz, Segmentation and classification of gait cycles, IEEE Trans. Neural Syst. Rehabil. Eng. 22 (2014) 946–952. [18] P. Bonato, T. D’Alessio, M. Knaflitz, A statistical method for the measurement of muscle activation intervals from surface myoelectric signal during gait, IEEE Trans. Biomed. Eng. 45 (1998) 287–299. [19] A.R. Den Otter, A.C. Geurts, T. Mulder, J. Duysens, Gait recovery is not associated with changes in the temporal patterning of muscle activity during treadmill walking in patients with post-stroke hemiparesis, Clin. Neurophysiol. 117 (1) (2006) 4–15. [20] A. Mengarelli, E. Maranesi, V. Barone, L. Burattini, S. Fioretti, F. Di Nardo, Evaluation of gender-related differences in co-contraction activity of shank muscles during gait, Conf. Proc. IEEE Eng. Med. Biol. Soc. (2015) 6066–6069. [21] F. Di Nardo, S. Fioretti, Emg-based analysis of treadmill and ground walking in distal leg muscles, IFMBE Proc. 41 (2014) 611–614, doi:http://dx.doi.org/ 10.1007/978-3-319-00846-2_151. [22] F. Di Nardo, E. Maranesi, A. Mengarelli, L. Burattini, S. Fioretti, Assessment of the ankle muscle co-contraction during normal gait: a surface electromyography study, J. Electromyogr. Kinesiol. 25 (2015) 347–354. [23] F. Di Nardo, A. Mengarelli, G. Ghetti, S. Fioretti, Statistical analysis of EMG signal acquired from tibialis anterior during gait, IFMBE Proc. 41 (2014) 619– 622, doi:http://dx.doi.org/10.1007/978-3-319-00846-2_153. [24] F. Di Nardo, A. Mengarelli, L. Burattini, E. Maranesi, V. Agostini, V. Nascimbeni, M. Knaflitz, S. Fioretti, Normative EMG patterns of ankle muscle co-contractions in school-age children during gait, Gait Posture 46 (2016) 161–166. [25] R.H. Gabel, R.A. Brand, The effects of signal conditioning on the statistical analyses of gait EMG, Electroenceph. Clin. Neurophysiol. 93 (1994) 188–201. [26] A. Nene, C. Byrne, H. Hermens, Is rectus femoris really part of the quadriceps? assessment of the rectus femoris function during gait in able-bodied adults, Gait Posture 10 (2004) 1–13. [27] A.R. den Otter, Speed related changes in muscle activity from normal to very slow speeds, Gait Posture 19 (2004) 270–278. [28] S. Ounpuu, P. DeLuca, K. Bell, R. Davis, Using surface electrodes for the evaluation of the rectus femoris, vastus medialis and the vastus lateralis muscles in children with cerebral palsy, Gait Posture 5 (1997) 211–221. [29] R.P. Grelsamer, J. McConnell, The Patella: A Team Approach, Aspen Publishers, Gaithersburg (MD), 1998.
Contents lists available at ScienceDirect
Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost
Full length article
Surface-EMG analysis for the quantification of thigh muscle dynamic co-contractions during normal gait Annachiara Strazzaa , Alessandro Mengarellia , Sandro Fiorettia , Laura Burattinia , Valentina Agostinib , Marco Knaflitzb , Francesco Di Nardoa,* a b
Department of Information Engineering, Università Politecnica delle Marche, 60131 Ancona, Italy Department of Electronics and Telecommunications, Politecnico di Torino, Torino, Italy
A R T I C L E I N F O
Article history: Received 27 May 2016 Received in revised form 26 October 2016 Accepted 1 November 2016 Keywords: sEMG Knee Statistical gait analysis Antagonist muscles Co-contraction
A B S T R A C T
The research purpose was to quantify the co-contraction patterns of quadriceps femoris (QF) vs. hamstring muscles during free walking, in terms of onset-offset muscular activation, excitation intensity, and occurrence frequency. Statistical gait analysis was performed on surface-EMG signals from vastus lateralis (VL), rectus femoris (RF), and medial hamstrings (MH), in 16315 strides walked by 30 healthy young adults. Results showed full superimpositions of MH with both VL and RF activity from terminal swing, 80 to 100% of gait cycle (GC), to the successive loading response (0–15% of GC), in around 90% of the considered strides. A further superimposition was detected during the push-off phase both between VL and MH activation intervals (38.6 12.8% to 44.1 9.6% of GC) in 21.9 13.6% of strides, and between RF and MH activation intervals (45.9 5.3% to 50.7 9.7 of GC) in 32.7 15.1% of strides. These findings led to identify three different co-contractions among QF and hamstring muscles during able-bodied walking: in early stance (in 90% of strides), in push-off (in 25–30% of strides) and in terminal swing (in 90% of strides). The co-contraction in terminal swing is the one with the highest levels of muscle excitation intensity. To our knowledge, this analysis represents the first attempt for quantification of QF/hamstring muscles co-contraction in young healthy subjects during normal gait, able to include the physiological variability of the phenomenon. ã 2016 Elsevier B.V. All rights reserved.
1. Introduction Position and complexity of knee joint suggest that it is relatively weak and susceptible to injury [1]. Thus, knee joint relies on muscles and ligaments to ensure stability. Main muscles carrying out this task are quadriceps femoris (QF) and hamstrings muscles. QF is the most direct source of extensor control [2]. During walking, contraction of quadriceps muscles represents the primary absorbing mechanism of impact during weight acceptance [2]. In stance, extensors act to decelerate knee flexion, while in swing they contribute to limb progression [2]. At knee joint, hamstrings are mainly involved in knee flexion control. During walking, hamstrings serve as a useful protective function in preventing knee
Abbreviations: F, flat foot contact; GC, gait cycle; H, heel contact; MH, medial hamstrings; P, push-off; QF, quadriceps femoris; RF, rectus femoris; S, swing; SGA, statistical gait analysis; VL, vastus lateralis. * Corresponding author at: Department of Information Engineering, Università Politecnica delle Marche, Via Brecce Bianche, 60131, Ancona, Italy. E-mail address: [email protected] (F. Di Nardo). http://dx.doi.org/10.1016/j.gaitpost.2016.11.003 0966-6362/ã 2016 Elsevier B.V. All rights reserved.
hyperextension in terminal swing and in following early stance [2]. QF and hamstring are antagonist muscle groups for knee joint during walking. Knee-joint stability is mainly preserved by co-contractions of antagonist muscle groups [3]. Co-contraction generally-held purpose is to augment ligament function in maintenance of joint stability, providing resistance to rotation at a joint and equalizing pressure distribution at joint surfaces [3]. Quantitative estimation of myoelectric activity of QF and hamstring muscles and of their concomitant recruitment could be very useful for analyzing kneejoint stability. Different studies assessed QF and hamstring muscles co-contractions during able-bodied walking, starting from the activation picture of rectus femoris, vastii and hamstring muscles [1,4–6]. Recent studies based on statistical gait analysis (SGA) of numerous strides/subject, reported that a single muscle showed a different number of activation intervals in different strides of the same walking [7–11]. Five different activation modalities were identified for both rectus femoris [8] and vastii [7], describing the large variability of muscular recruitment. This led to identifying muscular activities which are usually not reported in classic
A. Strazza et al. / Gait & Posture 51 (2017) 228–233
reference studies for EMG during gait [2,12], as for example VL activity in push-off phase [7,9]. Moreover, SGA-based studies also reported the occurrence frequency, defined as the frequency each muscle-activation occurs, quantified by the number of strides in which muscle is recruited with that specific activation modality [10]. This is a parameter hardly ever considered, because of low number of strides analyzed in traditional EMG studies. From this point of view, data on QF and hamstring muscles co-contractions reported by cited studies [1,4–6] are limited by the small number of gait cycles (GC) considered during an assessment session. The aim of the study was the quantitative assessment during able-bodied walking of the concomitant recruitment (i.e. cocontraction) of QF and hamstrings muscles in numerous (hundreds) strides per subject, in terms of variability of onset-offset muscular activation, excitation intensity, and occurrence frequency. Vastus lateralis (VL) and rectus femoris (RF) were analyzed as representative muscles for QF group. Medial hamstrings (MH) were considered as representative of hamstrings group. The technique of SGA [13,14] was performed to align with recent literature in handling numerous strides. 2. Materials and methods 2.1. Subjects Thirty healthy adults (15 females and 15 males) were recruited. Mean (SD) characteristics are: age = 23.8 1.9 years; height = 173 10 cm; weight = 63.3 12.4 kg; body mass index (BMI) = 20.8 2.1 kg m 2. Exclusion criteria included joints pain, neurological pathology, orthopedic surgery, abnormal gait or BMI 25. Participants signed informed consent. The research was undertaken in compliance with ethical principles of Helsinki Declaration and approved by institutional expert committee. 2.2. Signal acquisition and processing Signals were acquired (sampling rate:2 kHz; resolution:12 bit) and processed by the multichannel recording system, Step32 (Version PCI-32 ch2.0.1. DV, MedicalTechnology, Italy). Each subject was instrumented with foot-switches, knee electrogoniometers and sEMG electrodes on both lower limbs. Three foot-switches (size:11 11 0.5 mm; activation force:3 N) were attached beneath heel, first and fifth metatarsal heads of each foot. An electro-goniometer (accuracy:0.5 ) was attached to lateral side of each lower limb for measuring knee-joint angles in sagittal plane.; sEMG signals were detected with single-differential electrodes with fixed geometry (Ag/Ag-Cl disks, size:7 27 19 mm, electrode diameter:4 mm, interelectrode distance:12 mm,
229
gain:1000, high pass filter:10 Hz, input impedance >1.5 GV, CMRR> 126 dB, input referred noise 1 mVrms), and with variable geometry (Ag/Ag-Cl disks, minimum interelectrode distance:12 mm, gain:1000, high-pass filter:10 Hz, input impedance >1.5 GV, CMRR >126 dB, input referred noise 200 nVrms). sEMG signals were further amplified and low-pass filtered (450 Hz) by recording system.; Skin was shaved, cleansed with abrasive paste and wet with a soaked cloth. Electrodes were applied over VL, MH and RF, following SENIAM recommendations for electrode locationorientation over muscles with respect to tendons, motor points and fiber direction [15]. Participant set-up is shown in Fig. 1. Subjects were asked to walk barefoot overground for 5 min at natural pace, back and forth over a 12-meters-straight track, maintaining a steady speed. Natural pace was chosen because walking at comfortable speeds improves repeatability of EMG data [16]. Footswitch signals were debounced, converted to four levels, Heel contact (H), Flat foot contact (F), Push-off (P), Swing (S), and processed to segment and classify different GCs [17].; Electrogoniometric signals were low-pass filtered (FIR filter, 100 taps, cutoff frequency:15 Hz). Knee angles along with sequences and durations of gait phases derived by basographic signal were used by a multivariate statistical filter, to detect and discard outlier cycles like those relative to deceleration, reversing, and acceleration. sEMG signals were high-pass filtered (FIR filter, 100 taps, cutoff frequency:20 Hz) and processed by a double-threshold statistical detector, allowing a user-independent assessment of activation intervals [18]. This technique consists of selecting a first threshold z and observing m successive samples: if at least r0 (second threshold) out of successive m samples are above z, presence of the signal is acknowledged. Details are reported in [18]. Muscular co-contractions were quantified by assessing the overlapping period among activation intervals of considered muscles, in the very same strides [19]. Overlapping periods 30 ms were not considered in co-contraction computation, since a muscle activation 30 ms has no effect in controlling joint motion during gait [18]. 2.3. Statistical gait analysis Statistical gait analysis (SGA) performs a statistical characterization of gait analyzing spatial-temporal and sEMG-based parameters over numerous strides, during the same walking trial [7–11,13,14,20–24]. SGA is based on the observation that the number of muscle activations within a cycle is cycle dependent: it may vary from stride to stride. Therefore, averaging across strides should be performed only over those strides sharing the same number of activations, i.e. belonging to the same activation modality. Activation modality is defined as the number of times a
Fig. 1. Participant set-up.
230
A. Strazza et al. / Gait & Posture 51 (2017) 228–233
muscle activates during a single GC (n-activation modality consists of n activation intervals for the considered muscle, during a single GC). Mean activation intervals (normalized with respect to GC) for each activation modality are achieved, according to following steps. First, muscle activation intervals relative to each GC are identified, computing muscle onset/offset instants in temporal space [18], as previously described. Then, for each subject, gait cycles are grouped according to their activation modality, and onset-offset timings are averaged separately for each modality. Eventually, onset/offset time instants of each activation modality are averaged over the thirty subjects. Averaged onset/offset percentage time instants are normalized with respect to GC to provide mean activation intervals in percentage of GC. SGA was performed by Step32 system. SGA analysis of numerous strides allowed quantifying thigh co-contractions in terms of onset-offset muscular activation, quantized excitation intensity, and occurrence frequency (i.e. the number of times muscle is recruited with a specific activation modality during a single walking trial). Further details can be found in Supplementary material. 2.4. Assessment of excitation intensity A statistical representation of mean muscle excitation intensity over population was achieved as follows. Step32 provides the excitation intensity for each subject, quantized in three levels. In this study, the three intensity levels, considered in ascending order, were associated with the value of 1, 2 and 3, respectively. A global quantized mean signal over all subjects, ranging from 0 (no signal) to 90 (maximum value), was obtained for each percentage unit of GC, as the sum of intensity levels of every single mean signal. This global mean signal was subdivided in four intervals ranging from 0 to 12 (no muscle activation), from 13 to 30 (low-level activation), from 31 to 60 (medium-level activation), and from 61 to 90 (highlevel activation). Mean activation intervals are then represented by horizontal bars using a three-level grey scale for the excitation intensity; darker shadowing corresponds to increasing intensity of the mean signal, i.e. low-level (light grey), medium-level (dark grey), and high level (black) activation. The purpose of this quantization is mainly to identify the peak regions of myoelectric activity. 2.5. Statistics Data are reported as means standard deviation (SD). Lilliefors test was used to evaluate if each data vector had a normal distribution. The analysis of variance (ANOVA) test, followed by a multiple-comparison test, was used to compare normally distributed samples. To determine the power of an observed effect based on the sample size and parameter estimates derived from data set, the statistical power analysis was performed. Statistical significance was set at 5%. 3. Results Mean of 454 112 strides was considered for each subject. Durations of gait phases ware computed; H-phase lasts for 5.6 1.7% of GC, F-phase 28.9 6.0%, P-phase 23.4 4.8%, and Sphase 44.1 2.8%. Activation modalities for each muscle are reported in Fig. 2. Mean results are reported with data from right and left lower limb considered all together.; Different superimposition intervals between VL and RF activities were detected during GC. In strides characterized by RF double activation (Fig. 3A), superimpositions between RF and VL activation intervals were observed in early stance (1.9 5.4 to 23.3 14.1% of GC) and in terminal swing (81.4 2.9% to GC end). These superimpositions were detected in the two main activation modalities for VL. Also in
Fig. 2. Mean activation intervals vs. percentage of gait cycle for modalities with 1, 2, 3, 4 and 5 activations detected during walking for VL (panel A), RF (panel B), and MH (panel C). H, F, P and S phases are delimited by dashed vertical lines.
strides where RF adopted the 3- and 4-activation modalities, superimpositions were observed from terminal swing to the following early stance, for all VL activation modalities (from the beginning to 15.9 5.0% and from 85.6 3.3 to 99.6 0.8% of GC for RF triple activation, Fig. 3B; for RF 4-activation from the beginning to 14.3 3.3% and from 85.6 3.2% to the end of GC, Fig. 3C). In strides characterized by RF 4-activation (Fig. 2C), a further superimposition between VL and RF activation intervals was observed during mid-stance from 36.2 6.8% to 39.3 4.4% of GC, in 6.9 4.9% of considered strides. Occurrence frequencies of RF/VL co-contractions are reported in Fig. 3D. Different superimposition intervals were detected also between VL and MH activity. In strides with VL double activation (Fig. 4A), superimpositions with the two main MH activation intervals were observed from the beginning to 14.5 10.2% of GC) and from 81.5 3.3% to GC end. Also in strides where VL adopted 3(Fig. 4B) and 4-activation modality (Fig. 4C), VL/MH superimpositions were observed from terminal swing to the following early stance, for all MH activation modalities (from the beginning to 15.5 15.1% and from 81.8 3.8 to 99.9 0.1% of GC for VL triple activation; for VL 4-activation from the beginning to 14.8 14.2%
A. Strazza et al. / Gait & Posture 51 (2017) 228–233
231
and 4-activation (from the beginning of GC to 14.3 3.3%) modalities of MH and during swing, for 2-activation (from 80.6 10.7% to 98.2 4.8%), 3-activation (from 85.6 3.3% to the end of GC) and 4-activation (from 85.6 3.2% to 97.8 4.2%) modalities of MH. When RF increases to 3-activation modality, a co-contraction with MH occurs from 45.9 5.3% to 50.7 9.7 of GC (Fig. 5B). Occurrence frequencies of RF/MH co-contractions are reported in Fig. 5D. Further information on muscle co-contraction was given in terms of excitation intensity. Mean values over 30 subjects of muscle excitation intensity were computed following the procedure reported in “Assessment of excitation intensity” section and depicted by means of a three-level grey scale in Figs. 3–5. Percentage occurrence frequencies of co-contractions were compared. No significant difference was detected among occurrence frequencies of RF/VL (100%), VL/MH (90.3 13.3%), and RF/ MH (92.4 16.6%) co-contractions in early stance and in swing. RF/ MH co-contraction in push-off showed a significant higher occurrence frequency respect to the other push-off co-contractions. Results of statistical power analysis showed that number of participants (30) guarantees a power 0.90 for every difference tested in the study. 4. Discussion
Fig. 3. Mean values of VL activation intervals vs. percentage of gait cycle, detected in strides where RF shows 2-activation modality (panel A), 3-activation modality (panel B), and 4-activation modality (panel C). VL activation intervals are reported separately for modalities with 2 and 3 activations. Mean excitation intensity are represented by using a three-level grey scale; darker shadowing corresponds to increasing intensity of the mean signal, i.e. low-level (light grey), medium-level (dark grey), and high-level (black) activation. VL/RF co-contractions are highlighted by box. Mean (+SD) percentage occurrence frequency of each of the three different VL/RF co-contractions are reported in panel D. H, F, P and S phases are delimited by dashed grey vertical lines.
and from 81.5 4.1% to the end of GC). In strides with VL triple activation (Fig. 4B), a further VL/MH superimposition was observed from 38.6 12.8% to 44.1 9.6% of GC. Occurrence frequencies of VL/MH co-contractions are reported in Fig. 4D. Superimpositions between sEMG activity of RF and MH were also quantified. In strides where RF presented a double, triple, and quadruple activation (Fig. 5), a RF/MH superimposition was observed during early stance, for 2-activation (from 1.9 5.4% to 12.5 8.6%), 3-activation (from the beginning of GC to 12.8 6.3%)
In agreement with [7,8,11,20], this study showed that muscles adopt different activation modalities in different strides of the same walking. This finding was reported also for ankle muscles [21–23] and in children [9,24]. This means that a muscle could be recruited in a single stride with one, two, three or even more activities and in the following stride the number of activities could be different. Numerous strides allowed identifying five different activation modalities for each muscle, describing the large variability of muscular recruitment (Fig. 2). This leads to identifying muscular activities usually not reported in classic reference studies for EMG during gait [2,12], as VL activity in pushoff phase. The study of the whole variability of the three muscles allowed evaluating superimposition of muscle activity in all activation modalities, providing what the present study suggests to be the complete picture of co-contractions among VL, RF and MH. Optimal number of strides for providing a suitable average EMG profile was suggested to be 20, for natural walking speed [25]. Since averaged values were computed separately over five activation modalities for each muscle (Fig. 2), a total number of at least 100 strides per subject should be analyzed for a complete characterization of muscle recruitment variability. This matches also with our practical experience. RF-activity from terminal swing to the following loading response (Fig. 2) was detected in the nearly totally of strides. Main VL-activity occurred at the same relative percentage in GC as RF (Fig. 2), determining a superimposition in 100% of strides. These activities are recognized as typical activations of QF muscles as knee extensors. The superimposition during early stance could be intended as a synergic action of muscles for controlling weight bearing; this matches with [9]. Similarly, the synergic action of VL and RF in terminal swing is suggested to be present to assist knee extension and develop muscle tension for weight acceptance during loading response. However RF-signal, detected from terminal swing to the following loading response, could also be due to cross-talk from vastii [8,26]. High levels of muscle excitation intensity (Fig. 3) in these gait phases could support both hypotheses. MH was also recruited (Fig. 2) in the same percentage of GC where the above-cited activity of QF muscles was detected. This determined a superimposition of MH with both VL and RF activity in more than 90% of considered strides (Figs. 4–5). While it is
232
A. Strazza et al. / Gait & Posture 51 (2017) 228–233
Fig. 4. Mean values of MH activation intervals (black bars) vs. percentage of gait cycle, detected in strides where VL (grey bars) shows 2-activation modality (panel A), 3-activation modality (panel B), and 4-activation modality (panel C). MH activation intervals are reported separately for modalities with 2 and 3activations. Mean excitation intensity are represented by using a three-level grey scale; darker shadowing corresponds to increasing intensity of the mean signal, i.e. low-level (light grey), medium-level (dark grey), and high-level (black) activation. VL/MH cocontractions are highlighted by box Mean (+SD) percentage occurrence frequency of each of the three different VL/MH co-contractions are reported in panel D. H, F, P and S phases are delimited by dashed grey vertical lines.
reasonable indicating that mono-articular vastii muscles influence actions about the knee, it is real hard isolating the influence of a biarticular muscle (MH) to a single joint. Thus, interpretation of superimposition in early stance and swing needs to be deepened. However, superimposition in terminal swing could be reasonably
Fig. 5. Mean values of MH activation intervals (black bars) vs. percentage of gait cycle, detected in strides where RF (grey bars) shows 2-activation modality (panel A), 3-activation modality (panel B), and 4-activation modality (panel C). MH activation intervals are reported separately for modalities with 2 and 3 activations. Mean excitation intensity are represented by using a three-level grey scale; darker shadowing corresponds to increasing intensity of the mean signal, i.e. low-level (light grey), medium-level (dark grey), and high-level (black) activation. VL/MH cocontractions are highlighted by box. Mean (+SD) percentage occurrence frequency of each of the three different RF/MH co-contractions are reported in panel D. H, F, P and S phases are delimited by dashed grey vertical lines.
intended as an action of muscles across the same joint, the knee. It likely occurs in this gait phase to assist knee extension, developing muscle tension for weight acceptance during loading response [2,3]. This is the co-contraction with the highest level of muscle excitation intensity (Figs. 4–5). Moreover, it is involved in assisting the anterior cruciate ligament (ACL) to prevent excessive anterior
A. Strazza et al. / Gait & Posture 51 (2017) 228–233
tibial displacement and in protecting ACL itself from injury [1]. This supports the hypothesis of a regulatory role of the co-contraction between QF and hamstring muscles in providing knee joint stability [3]. RF and VL overlapped their activity also in push-off phase (Fig. 3C). RF and VL are likely working in synergy for modulating rapid knee flexion [7–9,27,28], although in only 8.14 15.7% of considered strides and with low levels of muscle excitation intensity (Fig. 3C). At the same time, VL is able to contribute to patella stabilization before entering pre-swing phase and RF participates to hip flexion [7–9,27,28]. Different tasks of muscles could probably explain the large variability of muscle activity and co-contraction in this phase. During push-off phase, uncommon and variable activities of VL and MH were detected (Fig. 2). VL is likely recruited to modulate rapid knee flexion and to stabilize the patella before entering preswing phase [29]. MH acts as hip extensor to propel body forward [2]. These activities implied the appearance of a superimposition between VL and MH (Fig. 4B). It was detected in a low percentage of strides (21.9 13.6%) and with low levels of muscle excitation intensity (Fig. 4B), and showed a large variability both intra- and inter-subjects, covering the entire push-off phase (around 30–50% of GC). This superimposition should not be considered as a real cocontraction, because VL and MH are working mainly on different joints. Differently, the superimposition between sEMG signals from RF and MH (Fig. 5B), detected in 32.7 15.1% of strides, could be intended as an actual co-contraction for the control of rapid knee extension and it could play a role also in stabilization of pelvis during body progression. The levels of excitation intensity appeared to be higher than those reported for VL/MH co-activation in the same phase (Fig. 5B vs. Fig. 4B). However, outcomes on this co-contraction should be confirmed and supported by further analyses that consider a direct study of muscle function via electrical stimulation. Differences in preferred walking speed across participants could introduce variability in muscle onset-offset times. Thus, walking speed could be considered a potential confound and this could be acknowledged as a limitation of the study. To our knowledge, this analysis represents the first attempt for quantification of QF/hamstring muscles co-contraction in young healthy subjects during normal gait. Three different co-contractions were identified: in early stance (in 90% of strides), in pushoff (in 25–30% of strides) and during swing (in 90% of strides). The contribution of the study consists in providing novel quantitative information on variability of thigh-muscle co-contractions, in terms of onset-offset muscular activation, excitation intensity and occurrence frequency. Thus, the present findings can be useful in clinical context and for designing future gait studies. Conflict of interest None. Acknowledgments The authors thank Dr. Giacomo G. Ghetti (Posture and Movement Analysis Laboratory, Italian National Institute of Health and Science on Aging (INRCA), Ancona, Italy) for the precious help he gave in clinical interpretation of data. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j. gaitpost.2016.11.003.
233
References [1] S.T. da Fonseca, P.L. Silva, J.M. Ocarino, R.B. Guimaraes, M.T. Oliveira, C.A. Lage, Analyses of dynamic co-contraction level in individuals with anterior cruciate ligament injury, J. Electromyogr. Kinesiol. 14 (2004) 239–247. [2] J. Perry, Gait Analysis–Normal and Pathological Function, Slack Incorporated, Thorofare (NJ), 1992. [3] R. Baratta, M. Solomonow, B.H. Zhou, D. Letson, R. Chuinard, R. D’Ambrosia, Muscular coactivation: the role of the antagonist musculature in maintaining knee stability, Am. J. Sports Med. 16 (1988) 113–122. [4] G. Frost, J. Dowling, K. Dyson, O. Bar-Or, Cocontraction in three age groups of children during treadmill locomotion, J. Electromyogr. Kinesiol. 7 (1997) 179–186. [5] T. Hortobagyi, S. Solnik, A. Gruber, P. Rider, K. Steinweg, J. Helseth, P. DeVita, Interaction between age and gait velocity in the amplitude and timing of antagonist muscle coactivation, Gait Posture 29 (2009) 558–564. [6] A. Schmitz, A. Silder, B. Heiderscheit, J. Mahoney, Differences in lower extremity muscular activation during walking between healthy older and young adults, J. Electromyogr. Kinesiol. 19 (2009) 1085–1091. [7] F. Di Nardo, E. Maranesi, A. Mengarelli, G. Ghetti, L. Burattini, S. Fioretti, Assessment of the variability of vastii myoelectric activity in young healthy females during walking: a statistical gait analysis, J. Electromyogr. Kinesiol. 25 (2015) 800–807. [8] F. Di Nardo, S. Fioretti, Statistical analysis of surface electromyographic signal for the assessment of rectus femoris modalities of activation during gait, J. Electromyogr. Kinesiol. 23 (2013) 56–61. [9] V. Agostini, A. Nascimbeni, A. Gaffuri, P. Imazio, M.G. Benedetti, M. Knaflitz, Normative emg activation patterns of school-age children during gait, Gait Posture 32 (2010) 285–289. [10] F. Di Nardo, V. Agostini, M. Knaflitz, A. Mengarelli, E. Maranesi, L. Burattini, S. Fioretti, The occurrence frequency: a suitable parameter for the evaluation of the myoelectric activity during walking, Conf. Proc. IEEE Eng. Med. Biol. Soc. (2015) 6070–6073. [11] F. Di Nardo, A. Mengarelli, E. Maranesi, L. Burattini, S. Fioretti, Gender differences in the myoelectric activity of lower limb muscles in young healthy subjects during walking, Biomed. Signal Process. Control 19 (2015) 14–22. [12] D.H. Sutherland, The evolution of clinical gait analysis part l: kinesiological EMG, Gait Posture 14 (2001) 61–70. [13] V. Agostini, M. Knaflitz, Statistical gait analysis, in: R.U. Acharya, F. Molinari, T. Tamura, D.S. Naidu, J.S. Suri (Eds.), Distributed Diagnosis and Home Healthcare, American Scientific Publishers, Stevenson Ranch, 2012. [14] V. Agostini, F. Lo Fermo, G. Massazza, M. Knaflitz, Does texting while walking really affect gait in young adults? J. Neuroeng. Rehabil. 12 (2015) 86. [15] B. Freriks, H.J. Hermens, C. Disselhorst-Klug, G. Rau, Development of recommendations for semg sensors and sensor placement procedures, J. Electromyogr. Kinesiol. 10 (2000) 361–374. [16] M.P. Kadaba, H.K. Ramakrishnan, M.E. Wootten, J. Gainey, G. Gorton, G.V.B. Cochran, Repeatability of kinematic, kinetic, and electromyographic data in normal adult gait, J. Orthop. Res. 6 (1989) 849–860. [17] V. Agostini, G. Balestra, M. Knaflitz, Segmentation and classification of gait cycles, IEEE Trans. Neural Syst. Rehabil. Eng. 22 (2014) 946–952. [18] P. Bonato, T. D’Alessio, M. Knaflitz, A statistical method for the measurement of muscle activation intervals from surface myoelectric signal during gait, IEEE Trans. Biomed. Eng. 45 (1998) 287–299. [19] A.R. Den Otter, A.C. Geurts, T. Mulder, J. Duysens, Gait recovery is not associated with changes in the temporal patterning of muscle activity during treadmill walking in patients with post-stroke hemiparesis, Clin. Neurophysiol. 117 (1) (2006) 4–15. [20] A. Mengarelli, E. Maranesi, V. Barone, L. Burattini, S. Fioretti, F. Di Nardo, Evaluation of gender-related differences in co-contraction activity of shank muscles during gait, Conf. Proc. IEEE Eng. Med. Biol. Soc. (2015) 6066–6069. [21] F. Di Nardo, S. Fioretti, Emg-based analysis of treadmill and ground walking in distal leg muscles, IFMBE Proc. 41 (2014) 611–614, doi:http://dx.doi.org/ 10.1007/978-3-319-00846-2_151. [22] F. Di Nardo, E. Maranesi, A. Mengarelli, L. Burattini, S. Fioretti, Assessment of the ankle muscle co-contraction during normal gait: a surface electromyography study, J. Electromyogr. Kinesiol. 25 (2015) 347–354. [23] F. Di Nardo, A. Mengarelli, G. Ghetti, S. Fioretti, Statistical analysis of EMG signal acquired from tibialis anterior during gait, IFMBE Proc. 41 (2014) 619– 622, doi:http://dx.doi.org/10.1007/978-3-319-00846-2_153. [24] F. Di Nardo, A. Mengarelli, L. Burattini, E. Maranesi, V. Agostini, V. Nascimbeni, M. Knaflitz, S. Fioretti, Normative EMG patterns of ankle muscle co-contractions in school-age children during gait, Gait Posture 46 (2016) 161–166. [25] R.H. Gabel, R.A. Brand, The effects of signal conditioning on the statistical analyses of gait EMG, Electroenceph. Clin. Neurophysiol. 93 (1994) 188–201. [26] A. Nene, C. Byrne, H. Hermens, Is rectus femoris really part of the quadriceps? assessment of the rectus femoris function during gait in able-bodied adults, Gait Posture 10 (2004) 1–13. [27] A.R. den Otter, Speed related changes in muscle activity from normal to very slow speeds, Gait Posture 19 (2004) 270–278. [28] S. Ounpuu, P. DeLuca, K. Bell, R. Davis, Using surface electrodes for the evaluation of the rectus femoris, vastus medialis and the vastus lateralis muscles in children with cerebral palsy, Gait Posture 5 (1997) 211–221. [29] R.P. Grelsamer, J. McConnell, The Patella: A Team Approach, Aspen Publishers, Gaithersburg (MD), 1998.