- Email: [email protected]
Statistical analysis of surface electromyographic signal for the assessment of rectus femoris modalities of activation during gait
Journal of Electromyography and Kinesiology 23 (2013) 56–61
Contents lists available at SciVerse ScienceDirect
Journal of Electromyography and Kinesiology journal homepage: www.elsevier.com/locate/jelekin
Statistical analysis of surface electromyographic signal for the assessment of rectus femoris modalities of activation during gait Francesco Di Nardo ⇑, Sandro Fioretti Department of Information Engineering, Università Politecnica delle Marche, Via Brecce Bianche, 60131 Ancona, Italy
a r t i c l e
i n f o
Article history: Received 28 October 2011 Received in revised form 26 June 2012 Accepted 27 June 2012
Keywords: EMG Quadriceps femoris Gait analysis Cross-talk
a b s t r a c t Aim of the present study was to identify the different modalities of activation of rectus femoris (RF) during gait at self-selected speed, by a statistical analysis of surface electromyographic signal from a large number (hundreds) of strides per subject. The analysis of ten healthy adults showed that RF is characterized by different activation modalities within different strides of the same walk. RF most recurrent modality (observed in 53 ± 6% of total strides) consists of three activations, at the beginning of gait cycle, around foot-off and in the terminal swing. Further two modalities of RF activation differ from the most recurrent one because of the lack of activity around foot-off (26 ± 6%) or the splitting into two (or three) small activations around stance-to-swing transition (17 ± 2%). Despite the large variability, our statistical analysis allowed to identify two patterns of activation that characterize completely the behavior of rectus femoris during gait. The first pattern, around stance-to-swing transition, can be monophasic, biphasic or triphasic and is necessary to control knee extension and hip flexion from pre-swing to initial swing. The second pattern, from terminal swing to following mid-stance, is likely due to the contribution of lowlevel RF activity and cross-talk from surrounding vastii. Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction The assessment of the activation patterns of the muscles of quadriceps femoris (QF) group, such as vastus medialis (VM), vastus lateralis (VL), and rectus femoris (RF), is recently receiving increasing attention in human gait analysis (Agostini et al., 2010; Barr et al., 2010; Nene et al., 2004). In particular the discussion focuses on the analysis of the signal taken from RF, the only biarticular muscle among the QF group, where different modalities of activation have been reported. Single activation of RF around stance-to-swing transition has been typically found utilizing finewire electromyography (fwEMG) (Barr et al., 2010; Nene et al., 2004; Perry, 1992). Further activations during loading response, like those reported for the vastii, have been found by means of both surface electromyography (sEMG) (den Otter, 2004; Ounpuu et al., 1997; Sienko Thomas et al., 1996) and fwEMG (Annaswamy et al., 1999) studies. These differences in detecting and then interpreting RF signal were often attributed to different methods (sEMG vs. fwEMG) of recording muscle activity (Barr et al., 2010; Nene et al., 2004), but no clear evidence has been found in literature. It is known that a single motor task can be performed in many ways with a similar end result (Bernstein, 1967). This motor redundancy suggests that the nervous system is capable of producing ⇑ Corresponding author. Fax: +39 0712204224. E-mail address: [email protected] (F. Di Nardo). 1050-6411/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jelekin.2012.06.011
different myoelectric activity patterns for a given movement, including gait task. This means that a single muscle can show a different number of activation intervals in different strides of the same walk, even when environmental and external conditions are fixed. It is, therefore, important to recognize the natural variability associated with muscle activity during free walking in order to improve the interpretation of EMG signals. This can be achieved by recording and analyzing the electromyographic signal over a large number of steps per subject, which allows to detect the different modalities (i.e. different number of activations within different strides of the same walk) of muscle activation. From this point of view, data on RF modalities of activation, reported by cited studies, are limited by the small number of gait cycles considered during an assessment session. A study over a larger number of steps should be considered. This goal, hard to be fulfilled by means of fine-wire EMG, because of its invasiveness and being not a completely painless technique, appears to be reachable by means of surface EMG. Thus, the aim of the present study was to identify the different modalities of activation of rectus femoris in healthy adults during gait at self-selected speed, by analyzing surface EMG signal from a large number (hundreds) of strides per subject. The study is based on the recent availability of robust techniques for the detection of muscle activation intervals, and specific tools for statistical gait analysis (Bonato et al., 1998; Balestra et al., 2002; Staude et al., 2001).
F. Di Nardo, S. Fioretti / Journal of Electromyography and Kinesiology 23 (2013) 56–61
57
2. Materials and methods 2.1. Subjects Ten healthy adult volunteers were recruited (age: 24.7 ± 0.6 years; height: 169 ± 2 cm; weight: 56.7 ± 1.5 kg; body mass index (BMI): 19.9 ± 0.3 kgm 2). Exclusion criteria included history of neurological pathology, orthopedic surgery within the previous year, acute or chronic knee pain or pathology, BMI P 25, or abnormal gait. Before the beginning of the test, all participants signed informed consent.
Fig. 2. Schematized representation of the path walked by the recruited subjects during the experiment; subjects walked barefoot over the floor for 6 min at their natural pace.
2.2. Recording system: signal acquisition and processing Signals were acquired by means of a multichannel recording system for statistical gait analysis (Step 32, DemItalia, Italy). Each subject was instrumented with foot-switches and sEMG probes on the left lower limb. Three foot-switches (size: 11 11 0.5 mm; activation force: 0.2 N) were attached beneath the heel, the first and the fifth metatarsal heads of each foot. Single differential (SD) sEMG probes with fixed geometry constituted by Ag-disks (manufacturer: DemItalia, size: 7 27 19 mm; interelectrode distance: 12 mm, gain: 1000, high-pass filter: 20 Hz) were attached over the vastus medialis (VM), vastus lateralis (VL), rectus femoris (RF2, reference position 2), and medial hamstrings (MH), following the SENIAM recommendations (Freriks et al., 1999) (Fig. 1). Two further probes were attached over the rectus femoris, +2 cm (RF1, position 1) and 2 cm (RF3, position 3) far from reference position RF2 in the direction of the line from
the anterior spina iliaca superior to the superior part of the patella, as depicted in Fig. 1. The electrodes RF1 and RF3 were positioned carefully, in order to avoid the possible EMG amplitude decrease occurring when the electrodes are placed over the innervation zone or myotendinous junction (DeLuca, 1997). After positioning the sensors, subjects were asked to walk barefoot over the floor for 6 min at their natural pace, following the path schematized in Fig. 2. Natural pace was chosen because walking at a self-selected speed improves the repeatability of EMG data (Kadaba et al., 1989), while variability increases when subjects are required to walk abnormally slow (Powers et al., 1996; Winter and Yack, 1987). Foot-switch signals were converted to four levels corresponding to Heel contact (H), Flat foot contact (F), Push off (P), Swing (S) and processed to segment and classify the different gait cycles. EMG signals were high-pass filtered and then processed by a doublethreshold statistical detector that allows to obtain, in an user-independent way, the muscle activation intervals (Bonato et al., 1998). 2.3. Statistical analysis In the present study only gait cycles consisting of the sequence of H–F–P–S and corresponding to straight walking were considered. For each subject and each muscle, mean (over the total strides) activation intervals and relative frequency of activation during walk were calculated for each activation modality. A statistical representation of mean amplitude of muscle activation over our population was achieved as follows. The statistical tool embedded in the recording system provides, for each subject, mean muscle activations in function of the percentage of gait cycle; the amplitude of the mean activations is quantized in three levels (Bonato et al., 1998). In the present study, the three amplitude levels, considered in ascending order, were associated with the value of 1, 2 and 3, respectively. A global quantized mean signal over all 10 subjects, ranging from 0 (no signal) to 30 (maximum value), was obtained for each percentage unit of the gait cycle, as the sum of the amplitude levels of every single mean signal provided by the recording system. This global mean signal was, in turn, subdivided in four intervals ranging from 0 to 4 (no muscle activation), from 5 to 10 (low-level activation), from 11 to 20 (medium-level activation), and from 21 to 30 (high-level activation). The purpose of this quantization is mainly to allow a statistical description, in function of the percentage of gait cycle, of the mean muscle activation intervals over the whole population and to find the signal peak of the myoelectric activity. 3. Results
Fig. 1. Frontal view of the electrode placements in left lower limb: VM = vastus medialis, VL = vastus lateralis, MH = medial hamstrings, RF1 = rectus femoris in position 1, RF2 = rectus femoris in position 2, RF3 = rectus femoris in position 3.
For each subject a mean (±SE) of 373 ± 10 strides has been considered. From the total of 3730 strides considered, 134 strides (3.6% of total strides) have been removed from the analysis because they did not follow the H–F–P–S foot-switch pattern. The
58
F. Di Nardo, S. Fioretti / Journal of Electromyography and Kinesiology 23 (2013) 56–61
Fig. 3. Mean (+SE) percentage frequency of each of the five different modalities of activation patterns detected during the walk in vastus medialis (VM), vastus lateralis (VL), medial hamstrings (MH), and rectus femoris in position 1 (RF1), position 2 (RF2) and position 3 (RF3), respectively. Mean and standard error refer to the whole population.
mean value (±SE) of 360 ± 9 strides was obtained for each subject, and the mean (±SE) H, F, P and S phases duration over our population was calculated. The H-phase lasts 5.0 ± 0.5% (percent of the gait cycle, mean and standard error), the F-phase 32.0 ± 1.9%, the P-phase 23.7 ± 1.9% and the S-phase 39.2 ± 0.9%. Statistical analysis of the myoelectric signal in each single subject put in evidence that muscles show a different number of activation intervals in different strides of the same walk; five modalities of activation were observed, consisting of one, two, three, four and five muscle activation intervals, respectively. Fig. 3 shows, for each muscle, and for the whole subject population, the mean (+SE) percentage frequency for each of the five different modalities of activation patterns detected during walk. Fig. 4 shows the mean, over whole subject population, (+SE) activation intervals of the actual signal recorded from RF in reference position (RF2), for each of the five different modalities of activation. Details of mean (±SE) activation intervals of all considered muscles in their most recurrent modality of activation, expressed as the timing, in the percentage of gait cycle (GC), of signal onset and offset are reported in Table 1. A further representation of mean muscle
activation was given in terms of three amplitude levels. In this pictorial representation of data, mean values over ten subjects of muscle activation intervals were calculated following the procedure reported in Section 2.3 considering first the most recurrent (Fig. 5) and then the following less recurrent modalities of muscle activation for each subject (Fig. 6). Activation intervals are represented by horizontal bars whose area is shadowed by a three-level grey scale; darker shadowing corresponds to increasing amplitude of the mean signal, according to the definition of low-level (light grey), medium-level (dark grey), and high-level (black) activation given in Section 2.3. The most recurrent pattern of activation consists of two activations for VM (observed in 61 ± 7% of total strides) and VL (62 ± 6%), from heel strike to mid-stance and from terminal swing to the following initial contact; high-level activity (black area in Fig. 5) was measured at the beginning of gait cycle (0–4% for VM and 0–7% for VL) and in the terminal swing (90–100% for both VM and VL). For RF, in the reference position RF2, the most recurrent pattern of activation during gait cycle consists of three activations (observed in 53 ± 6% of total strides), at the beginning of gait cycle,
Fig. 4. Mean (+SE) activation intervals vs. percentage of gait cycle for five different modalities with 1, 2, 3, 4 and 5 activations, respectively, detected during the walk in rectus femoris in reference position (RF2). Mean percentage frequency of each modality of activation is reported on the right-hand side of the plot. Mean and standard error refer to the whole population. The heel contact, flat foot contact, push off and swing phases are delimited by dashed vertical lines.
59
F. Di Nardo, S. Fioretti / Journal of Electromyography and Kinesiology 23 (2013) 56–61 Table 1 Activation intervals of considered muscles in their most recurrent modality of activation. Muscle
VM VL RF1 RF2 RF3 MH
First activation (% gait cycle)
Second activation (% gait cycle)
Third activation (% gait cycle)
ON
OFF
ON
OFF
ON
OFF
0.16 ± 0.09 0.13 ± 0.11 0.46 ± 0.26 0.19 ± 0.01 0.34 ± 0.21 1.32 ± 0.47
16.6 ± 2.3 15.1 ± 2.7 14.7 ± 1.4 14.4 ± 1.1 15.0 ± 1.4 10.8 ± 1.0
59.9 ± 1.2 59.5 ± 1.0 59.3 ± 1.1 45.6 ± 2.3
85.8 ± 0.8 85.5 ± 0.7 87.8 ± 0.6 87.5 ± 0.6 86.8 ± 0.5 73.9 ± 0.8
99.5 ± 0.3 99.7 ± 0.2 99.7 ± 0.2 99.8 ± 0.1 99.9 ± 0.1 99.7 ± 0.1
46.3 ± 1.7 47.0 ± 0.8 45.9 ± 1.2 35.0 ± 2.0
Activation intervals are expressed as the timing, in the percentage of gait cycle (GC), of signal onset and offset for vastus medialis (VM), vastus lateralis (VL), medial hamstrings (MH), and rectus femoris in position 1 (RF1), position 2 (RF2) and position 3 (RF3). Values are expressed as mean ± inter-subject SE.
Fig. 5. Quantized average patterns of muscle activation intervals vs. percentage of gait cycle (GC) for vastus medialis (VM), vastus lateralis (VL), medial hamstrings (MH), and rectus femoris in position 1 (RF1), position 2 (RF2) and position 3 (RF3), in their most recurrent activation modality. Mean activation intervals are represented by horizontal bars using a three-level grey scale; darker shadowing corresponds to increasing amplitude of the mean signal, i.e. low-level (light grey), medium-level (dark grey), and highlevel (black) activation. The heel contact, flat foot contact, push off and swing phases are delimited by dashed vertical lines. All average data refer to the whole tested population.
around foot-off and in the terminal swing (Fig. 4, Table 1). The second most recurrent modality of activation for RF2 is similar to the 3-activation modality but with no activation in the transition between stance and swing. This 2-activation modality was observed in 26 ± 6% of total strides. In a further 14 ± 2% of total strides, four
activations were observed: at the beginning of gait cycle, around heel-off, around foot-off and in the terminal swing. The 5-activation modality, where three activations were observed around stance-to swing-transition, was relative to 3 ± 1% of total strides. The remaining 4 ± 2% of total strides was defined by a single
Fig. 6. Quantized average patterns of muscle activation intervals vs. percentage of gait cycle (GC) in the percentage of total strides (14 ± 2%) where four activation intervals were detected for rectus femoris in reference position (RF2). Mean activation intervals are represented by horizontal bars using a three-level grey scale; darker shadowing corresponds to increasing amplitude of the mean signal, i.e. low-level (light grey), medium-level (dark grey), and high-level (black) activation. The heel contact, flat foot contact, push off and swing phases are delimited by dashed vertical lines. All average data refer to the whole tested population.
60
F. Di Nardo, S. Fioretti / Journal of Electromyography and Kinesiology 23 (2013) 56–61
activation around foot-off (Fig. 4). Despite no significant differences in the total length of activation interval, in 3- (Fig. 5), 4(Fig. 6) and 5-activation modalities, the RF high-level activity (black area) measured at the beginning of gait cycle and in the terminal swing, increased approaching the vastii from RF1 to RF2, and then from RF2 to RF3. On the other hand, no significant change (or, in case, a mild decrease) in amplitude levels of RF activity around foot-off and/or around heel-off has been detected from RF1 to RF2 and then RF3. Comparable results were found for the 2-activation modality, in the transition between swing and stance.
4. Discussion In the present study, EMG signals were recorded from quadriceps femoris muscle group during gait in ten healthy adult subjects, in order to determine the picture of the activation modalities of rectus femoris (RF), based on a statistical analysis of a large number (approximately 350) of strides for each subject. The muscle activation intervals, detected in each subject at self-selected walking speed, followed roughly the typical pattern reported for each muscle during gait. However, the statistical analysis put into evidence that QF muscles showed different modalities in the number of activations and in the timing of signal onset and offset (Figs. 3 and 4). This suggests that it is worth considering not only the activation patterns of each muscle, but also how frequently they are observed. On average, the most recurrent activation pattern of the vastii was observed in a high percentage of the strides of the left lower limb; 61 ± 7% for VM and 62 ± 6% for VL (Fig. 3). It consists of two activations, from heel strike to mid-stance and from terminal swing to the following initial contact (Table 1, Fig. 5). The 2-activation modality found for the vastii is expected (Perry, 1992; Sutherland, 2001; Winter, 1991) and is required to generate tension in terminal swing in preparation for weight bearing at initial contact and to control knee flexion during weight acceptance. The contemporary recruitment of VM and VL found in this study, is consistent with previous results in healthy subjects (Cowan et al., 2001; Karst and Willett, 1995; Powers et al., 1996). The most recurrent modality of activation for rectus femoris in reference position, RF2, (observed in 53 ± 6% of total strides) consists of three activations, at the beginning of gait cycle, around stance-to-swing transition and in the terminal swing (Table 1, Fig. 4). In a further percentage (17 ± 2%) of strides, the pattern is similar, except for the activation around stance-to-swing transition that splits into two (or three) small activations, between heel-off and foot-off. In the remaining 26 ± 6% of total strides, the activity at the transient between stance and swing lacks, though activity in the swing-to-stance transition is still present. Despite the large variability in the modality of RF activation (Fig. 4), our statistical analysis points out the recurrence of two patterns of activation in the nearly totality of the strides and suggests that these patterns characterize completely the behavior of the rectus femoris during gait at self-selected speed. The first pattern of activity occurs around stance-to-swing transition and it can be monophasic (57 ± 5% of total strides), biphasic (14 ± 2%) or triphasic (3 ± 1%). This RF activity around the middle of gait cycle, reported by both sEMG-based (den Otter, 2004; Ounpuu et al., 1997; Sienko Thomas et al., 1996; Sutherland, 2001), and fine-wire-based studies (Barr et al., 2010; Nene et al., 2004; Perry, 1992), has been interpreted as active participation in rapid knee extension or hip flexion, or both (Conrad et al., 1986). A lack of this activation is observed in the 26 ± 6% of total strides. No RF activity in 25% of total strides during stance-to-swing transition is also reported in a recent study on school-age children (Agostini et al., 2010). This is likely due to those strides where self-selected speed was
temporary reduced, as reported by Nene et al. (2004). The second pattern of activity occurs from terminal swing to following midstance and is present in the nearly totality of the strides (96 ± 6%). The presence of RF activity during this portion of the gait cycle has been reported by means of sEMG analysis (den Otter, 2004; Ounpuu et al., 1997; Sienko Thomas et al., 1996). The activity during early stance phase has been related to RF eccentric action to assist the vastii to control weight bearing (Shiavi, 1985; Shiavi et al., 1987; Murray, 1984); the activity in terminal swing is supposed to be present to assist knee extension and helps developing muscle tension for weight acceptance during loading response to control knee flexion (Ounpuu et al., 1997). Otherwise, Perry (1989) rejected the role of RF in participating with the vastii to provide knee extension stability during limb loading and early midstance, resting on the lack of RF from terminal swing to following mid-stance observed utilizing fine-wire EMG. This absence of signal has been reported also by more recent fine-wire studies (Barr et al., 2010; Nene et al., 2004; Perry, 1992). An explanation to this disagreement has been suggested by the observation that the apparent RF activity commonly recorded during loading response using sEMG is actually the result of cross-talk from the surrounding vastii (Nene et al., 2004; Winter et al., 1994). Annaswamy et al. (1999), however, rejected the cross-talk hypothesis, reporting three activations of RF from fine-wires data. The issue is still far from being solved. Our statistical analysis showed that, on average, the high-level activation (i.e. the peak of EMG activity) of both VM and VL occurred at approximately 0–5% and 90–100% of the gait cycle. In the 96 ± 6% of total strides considering 2-, 3-, 4- and 5activation modalities all together, the high-level activation of RF also occurred at the same relative percentage time in the gait cycle as the vastii peak (see Figs. 5 and 6 for 3- and 4-activation modality, respectively); this seems to indicate the cross-talk from the vastii as the most likely origin of this phase of the RF surface EMG. However, just because surface EMG signals recorded from different muscles are modulated similarly, it does not necessarily imply cross-talk. Thus, the present study attempted a rough evaluation of this phenomenon, by means of the analysis of the variation of the mean amplitude of RF signal detected in three consecutive positions, progressively approaching the vastii (from position RF1 to RF3, Fig. 1). The increase of RF signal at the beginning of gait cycle and in the terminal swing, and no variation from pre-swing to the beginning of the swing phase (found in the 93 ± 5% of total strides), moving from position RF1 to RF3, seems to support the cross-talk hypothesis (see Figs. 5 and 6 for 3- and 4-activation modality, respectively). However, data available in the present study do not allow to argue that RF activity measured with surface electrodes during this part of the cycle is entirely due to cross-talk activity. Thus, the possibility that RF may be active albeit at a much lower level of activity and that cross-talk simply overrates the contribution of the RF cannot be refuted. A reliable investigation on cross-talk should be performed by triggering surface EMGs in the RF muscle with the firing patterns of motor units identified, with intramuscular electrodes (Merletti and Farina, 2009), in the vasti muscles, but this is beyond the goal of the present study. Hypothesis of possible cross-talk from the hamstrings, has been excluded by the observation that the low-level MH activities in the first 10% and high-level MH activities between 80% and 90% of gait cycle do not match with high-level RF activities and no RF activities, respectively, detected in the same percentages of gait cycle (Fig. 5). Some of the results reported in the present study were achieved by interpreting quantized sEMG signals. The quantization process could cause an underestimation of the low-level EMG activity; however, this possible underestimation does not affect the study, since the reported conclusions were mostly based on the analysis of the high-level activity (signal peak) of the muscles of quadriceps group.
F. Di Nardo, S. Fioretti / Journal of Electromyography and Kinesiology 23 (2013) 56–61
5. Conclusions The processing of the surface electromyographic signal, recorded on rectus femoris (RF) over a large number (approximately 350) of strides for each tested subject, points out that RF shows different activation modalities within different strides of the same walk. The statistical analysis of these different activation modalities suggests that, during gait at self-selected speed, RF is recruited with an activation modality characterized by two patterns of activity. The first pattern, observed around stance-to-swing transition, can occur in monophasic, biphasic or triphasic shape and it is necessary to control knee extension and hip flexion from pre-swing to initial swing. The second pattern of activity occurs from terminal swing to following mid-stance in the nearly totality of the strides. The increased RF signal activity detected around the swing-tostance transition progressively approaching the vastii, suggested cross-talk from the surrounding vastii as one of the most likely origins of this phase of the RF surface EMG. However, the possibility that RF may be active albeit at a much lower level of activity and that cross-talk eventually overrate the actual contribution of the RF cannot be refuted. References Agostini V, Nascimbeni A, Gaffuri A, Imazio P, Benedetti MG, Knaflitz M. Normative EMG activation patterns of school-age children during gait. Gait Posture 2010;32(3):285–9. Annaswamy TM, Giddings CJ, Croce UD, Kerrigan DC. Rectus femoris: its role in normal gait. Arch Phys Med Rehabil 1999;80:930–4. Balestra G, Knaflitz M, Molinari F. Principles of statistical gait analysis. XIV ISEK Congress, Vienna, Austria June 22–25, 2002. p. 63–64. Barr KM, Miller AL, Chapin KB. Surface electromyography does not accurately reflect rectus femoris activity during gait: impact of speed and crouch on vasti-torectus cross-talk. Gait Posture 2010;32:363–8. Bernstein N. Coordination and regulation of movements. New York: Pergamon; 1967. Bonato P, D’Alessio T, Knaflitz M. A statistical method for the measurement of muscle activation intervals from surface myoelectric signal during gait. IEEE Trans Biomed Eng 1998;45:287–99. Conrad B, Meinck HM, Benecke R. Motor patterns in human gait: adaptation to different modes of progression. In: Bles W, Brandt W, editors. Disorders of posture and gait. London: Elsevier; 1986. p. 53–67. Cowan SM, Bennell KL, Hodges PW, Crossley KM, McConnell J. Delayed onset of electromyographic activity of vastus medialis obliquus relative to vastus lateralis in subjects with patellofemoral pain syndrome. Arch Phys Med Rehabil 2001;82(2):183–9. DeLuca CJ. The use of surface electromyography in biomechanics. J Appl Biomech 1997;13:135–63. den Otter AR. Speed related changes in muscle activity from normal to very slow speeds. Gait Posture 2004;19:270–8. Freriks B, Hermens HJ, Disselhorst-Klug C, Rau G. The recommendations for sensors and sensor placement procedures for surface electromyography. In: Hermens HJ, Freriks B, Merletti R, Stegeman D, Bok J, Rau G, et al., editors. European recommendations for surface electromyography. Results of the SENIAM project. Enschede: Roessingh Research and Development; 1999. Kadaba MP, Ramakrishnan HK, Wootten ME, Gainey J, Gorton G, Cochran GVB. Repeatability of kinematic, kinetic, and electromyographic data in normal adult gait. J Orthop Res 1989;6:849–60. Karst GM, Willett GM. Onset timing of electromyographic activity in the vastus medialis oblique and vastus lateralis muscles in subjects with and without patellofemoral pain syndrome. Phys Ther 1995;75:813–23. Merletti R, Farina D. Analysis of intramuscular electromyogram signals. Philos Transact A Math Phys Eng Sci 2009;367:357–68. Murray MP. Kinematic & EMG patterns during slow, free, and fast walking. J Orthop Res 1984;2(3):272–80.
61
Nene A, Byrne C, Hermens H. Is rectus femoris really part of the quadriceps? Assessment of the rectus femoris function during gait in able-bodied adults. Gait Posture 2004;10:1–13. Ounpuu S, DeLuca P, Bell K, Davis R. Using surface electrodes for the evaluation of the rectus femoris, vastus medialis and the vastus lateralis muscles in children with cerebral palsy. Gait Posture 1997;5:211–6. Perry J. Gait analysis - normal and pathological function. USA: Slack Inc.; 1992. Perry J, Gronley JK, Bolanger EL. Functional role of the rectus femoris in gait. In: Proceedings of the 35th Annual Meeting of the Orthopaedic Research Society, 1989, Las Vegas USA. Powers CM, Landel RF, Perry J. Timing and intensity of vastus muscle activity during functional activities in subjects with and without patellofemoral pain. Phys Ther 1996;76:946–55. Shiavi R. Electromyographic patterns in adult locomotion: a comprehensive review. J Rehabil Res Dev 1985;22(3):85–98. Shiavi R, Bugle HJ, Limbird T. Electromyographic gait assessment, part 1: adult EMG profiles and walking speed. J Rehabil Res 1987;24(2):13–23. Sienko Thomas S, Moore C, Kelp Lenane C, Norris C. Simulated gait patterns:the resulting effects on gait parameters, dynamic electromyography, joint moments, and physiological cost index. Gait Posture 1996;4:100–7. Staude G, Flachenecker C, Daumer M, Wolf W. Onset detection in surface electromyographic signals: a systematic comparison of methods. EURASIP J Appl Signal Process 2001;2:67–81. Sutherland DH. The evolution of clinical gait analysis part l: kinesiological EMG. Gait Posture 2001;14:61–70. Winter DA. The biomechanics and motor control of human gait: normal elderly and pathological. 2nd ed. Ontario: Waterloo Biomechanics; 1991. Winter DA, Fuglevand AJ, Archer SE. Cross-talk in surface electromyography: theoretical and practical estimates. J Electromyogr Kinesiol 1994:415–26. Winter DA, Yack HJ. EMG profiles during normal human walking: stride to stride and inter-subject variability. Electroencephalogr Clin Neurophysiol 1987;67:402–11.
Francesco Di Nardo received his Ph.D. in Artificial intelligence systems from Università Politecnica delle Marche, Ancona, Italy in 2005. His main research activity involves the development and the clinical application (type-2 diabetes, insulin resistance and hypertension) of mathematical models of metabolic and endocrine systems. In this field, he published several papers in peer review journals and participated in National research projects. Recently, he is addressing his research in the field of movement analysis for motor rehabilitation, with particular involvement in the acquisition and the processing of surface electromyography signal to assess the muscular function during gait task.
Sandro Fioretti graduated in 1979 in Electronic Engineering at Ancona University and presently is Associate Professor in Bioengineering at the Department of Information Engineering – Università Politecnica delle Marche – Ancona. He teaches Movement Biomechanics and Bioengineering of Motor Rehabilitation at the Biomedical Engineering course of the same University. His main research interests are in the field of human movement analysis and its related fields such as: stereophotogrammetry, linear and nonlinear filtering, joint kinematics, analysis and identification of postural control, static and perturbed posturography, gait analysis, dynamic electromyography. He participated in various European and National research projects in the field of movement analysis for motor rehabilitation. He is author of numerous scientific publications in international journals, books and congress proceedings.
Contents lists available at SciVerse ScienceDirect
Journal of Electromyography and Kinesiology journal homepage: www.elsevier.com/locate/jelekin
Statistical analysis of surface electromyographic signal for the assessment of rectus femoris modalities of activation during gait Francesco Di Nardo ⇑, Sandro Fioretti Department of Information Engineering, Università Politecnica delle Marche, Via Brecce Bianche, 60131 Ancona, Italy
a r t i c l e
i n f o
Article history: Received 28 October 2011 Received in revised form 26 June 2012 Accepted 27 June 2012
Keywords: EMG Quadriceps femoris Gait analysis Cross-talk
a b s t r a c t Aim of the present study was to identify the different modalities of activation of rectus femoris (RF) during gait at self-selected speed, by a statistical analysis of surface electromyographic signal from a large number (hundreds) of strides per subject. The analysis of ten healthy adults showed that RF is characterized by different activation modalities within different strides of the same walk. RF most recurrent modality (observed in 53 ± 6% of total strides) consists of three activations, at the beginning of gait cycle, around foot-off and in the terminal swing. Further two modalities of RF activation differ from the most recurrent one because of the lack of activity around foot-off (26 ± 6%) or the splitting into two (or three) small activations around stance-to-swing transition (17 ± 2%). Despite the large variability, our statistical analysis allowed to identify two patterns of activation that characterize completely the behavior of rectus femoris during gait. The first pattern, around stance-to-swing transition, can be monophasic, biphasic or triphasic and is necessary to control knee extension and hip flexion from pre-swing to initial swing. The second pattern, from terminal swing to following mid-stance, is likely due to the contribution of lowlevel RF activity and cross-talk from surrounding vastii. Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction The assessment of the activation patterns of the muscles of quadriceps femoris (QF) group, such as vastus medialis (VM), vastus lateralis (VL), and rectus femoris (RF), is recently receiving increasing attention in human gait analysis (Agostini et al., 2010; Barr et al., 2010; Nene et al., 2004). In particular the discussion focuses on the analysis of the signal taken from RF, the only biarticular muscle among the QF group, where different modalities of activation have been reported. Single activation of RF around stance-to-swing transition has been typically found utilizing finewire electromyography (fwEMG) (Barr et al., 2010; Nene et al., 2004; Perry, 1992). Further activations during loading response, like those reported for the vastii, have been found by means of both surface electromyography (sEMG) (den Otter, 2004; Ounpuu et al., 1997; Sienko Thomas et al., 1996) and fwEMG (Annaswamy et al., 1999) studies. These differences in detecting and then interpreting RF signal were often attributed to different methods (sEMG vs. fwEMG) of recording muscle activity (Barr et al., 2010; Nene et al., 2004), but no clear evidence has been found in literature. It is known that a single motor task can be performed in many ways with a similar end result (Bernstein, 1967). This motor redundancy suggests that the nervous system is capable of producing ⇑ Corresponding author. Fax: +39 0712204224. E-mail address: [email protected] (F. Di Nardo). 1050-6411/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jelekin.2012.06.011
different myoelectric activity patterns for a given movement, including gait task. This means that a single muscle can show a different number of activation intervals in different strides of the same walk, even when environmental and external conditions are fixed. It is, therefore, important to recognize the natural variability associated with muscle activity during free walking in order to improve the interpretation of EMG signals. This can be achieved by recording and analyzing the electromyographic signal over a large number of steps per subject, which allows to detect the different modalities (i.e. different number of activations within different strides of the same walk) of muscle activation. From this point of view, data on RF modalities of activation, reported by cited studies, are limited by the small number of gait cycles considered during an assessment session. A study over a larger number of steps should be considered. This goal, hard to be fulfilled by means of fine-wire EMG, because of its invasiveness and being not a completely painless technique, appears to be reachable by means of surface EMG. Thus, the aim of the present study was to identify the different modalities of activation of rectus femoris in healthy adults during gait at self-selected speed, by analyzing surface EMG signal from a large number (hundreds) of strides per subject. The study is based on the recent availability of robust techniques for the detection of muscle activation intervals, and specific tools for statistical gait analysis (Bonato et al., 1998; Balestra et al., 2002; Staude et al., 2001).
F. Di Nardo, S. Fioretti / Journal of Electromyography and Kinesiology 23 (2013) 56–61
57
2. Materials and methods 2.1. Subjects Ten healthy adult volunteers were recruited (age: 24.7 ± 0.6 years; height: 169 ± 2 cm; weight: 56.7 ± 1.5 kg; body mass index (BMI): 19.9 ± 0.3 kgm 2). Exclusion criteria included history of neurological pathology, orthopedic surgery within the previous year, acute or chronic knee pain or pathology, BMI P 25, or abnormal gait. Before the beginning of the test, all participants signed informed consent.
Fig. 2. Schematized representation of the path walked by the recruited subjects during the experiment; subjects walked barefoot over the floor for 6 min at their natural pace.
2.2. Recording system: signal acquisition and processing Signals were acquired by means of a multichannel recording system for statistical gait analysis (Step 32, DemItalia, Italy). Each subject was instrumented with foot-switches and sEMG probes on the left lower limb. Three foot-switches (size: 11 11 0.5 mm; activation force: 0.2 N) were attached beneath the heel, the first and the fifth metatarsal heads of each foot. Single differential (SD) sEMG probes with fixed geometry constituted by Ag-disks (manufacturer: DemItalia, size: 7 27 19 mm; interelectrode distance: 12 mm, gain: 1000, high-pass filter: 20 Hz) were attached over the vastus medialis (VM), vastus lateralis (VL), rectus femoris (RF2, reference position 2), and medial hamstrings (MH), following the SENIAM recommendations (Freriks et al., 1999) (Fig. 1). Two further probes were attached over the rectus femoris, +2 cm (RF1, position 1) and 2 cm (RF3, position 3) far from reference position RF2 in the direction of the line from
the anterior spina iliaca superior to the superior part of the patella, as depicted in Fig. 1. The electrodes RF1 and RF3 were positioned carefully, in order to avoid the possible EMG amplitude decrease occurring when the electrodes are placed over the innervation zone or myotendinous junction (DeLuca, 1997). After positioning the sensors, subjects were asked to walk barefoot over the floor for 6 min at their natural pace, following the path schematized in Fig. 2. Natural pace was chosen because walking at a self-selected speed improves the repeatability of EMG data (Kadaba et al., 1989), while variability increases when subjects are required to walk abnormally slow (Powers et al., 1996; Winter and Yack, 1987). Foot-switch signals were converted to four levels corresponding to Heel contact (H), Flat foot contact (F), Push off (P), Swing (S) and processed to segment and classify the different gait cycles. EMG signals were high-pass filtered and then processed by a doublethreshold statistical detector that allows to obtain, in an user-independent way, the muscle activation intervals (Bonato et al., 1998). 2.3. Statistical analysis In the present study only gait cycles consisting of the sequence of H–F–P–S and corresponding to straight walking were considered. For each subject and each muscle, mean (over the total strides) activation intervals and relative frequency of activation during walk were calculated for each activation modality. A statistical representation of mean amplitude of muscle activation over our population was achieved as follows. The statistical tool embedded in the recording system provides, for each subject, mean muscle activations in function of the percentage of gait cycle; the amplitude of the mean activations is quantized in three levels (Bonato et al., 1998). In the present study, the three amplitude levels, considered in ascending order, were associated with the value of 1, 2 and 3, respectively. A global quantized mean signal over all 10 subjects, ranging from 0 (no signal) to 30 (maximum value), was obtained for each percentage unit of the gait cycle, as the sum of the amplitude levels of every single mean signal provided by the recording system. This global mean signal was, in turn, subdivided in four intervals ranging from 0 to 4 (no muscle activation), from 5 to 10 (low-level activation), from 11 to 20 (medium-level activation), and from 21 to 30 (high-level activation). The purpose of this quantization is mainly to allow a statistical description, in function of the percentage of gait cycle, of the mean muscle activation intervals over the whole population and to find the signal peak of the myoelectric activity. 3. Results
Fig. 1. Frontal view of the electrode placements in left lower limb: VM = vastus medialis, VL = vastus lateralis, MH = medial hamstrings, RF1 = rectus femoris in position 1, RF2 = rectus femoris in position 2, RF3 = rectus femoris in position 3.
For each subject a mean (±SE) of 373 ± 10 strides has been considered. From the total of 3730 strides considered, 134 strides (3.6% of total strides) have been removed from the analysis because they did not follow the H–F–P–S foot-switch pattern. The
58
F. Di Nardo, S. Fioretti / Journal of Electromyography and Kinesiology 23 (2013) 56–61
Fig. 3. Mean (+SE) percentage frequency of each of the five different modalities of activation patterns detected during the walk in vastus medialis (VM), vastus lateralis (VL), medial hamstrings (MH), and rectus femoris in position 1 (RF1), position 2 (RF2) and position 3 (RF3), respectively. Mean and standard error refer to the whole population.
mean value (±SE) of 360 ± 9 strides was obtained for each subject, and the mean (±SE) H, F, P and S phases duration over our population was calculated. The H-phase lasts 5.0 ± 0.5% (percent of the gait cycle, mean and standard error), the F-phase 32.0 ± 1.9%, the P-phase 23.7 ± 1.9% and the S-phase 39.2 ± 0.9%. Statistical analysis of the myoelectric signal in each single subject put in evidence that muscles show a different number of activation intervals in different strides of the same walk; five modalities of activation were observed, consisting of one, two, three, four and five muscle activation intervals, respectively. Fig. 3 shows, for each muscle, and for the whole subject population, the mean (+SE) percentage frequency for each of the five different modalities of activation patterns detected during walk. Fig. 4 shows the mean, over whole subject population, (+SE) activation intervals of the actual signal recorded from RF in reference position (RF2), for each of the five different modalities of activation. Details of mean (±SE) activation intervals of all considered muscles in their most recurrent modality of activation, expressed as the timing, in the percentage of gait cycle (GC), of signal onset and offset are reported in Table 1. A further representation of mean muscle
activation was given in terms of three amplitude levels. In this pictorial representation of data, mean values over ten subjects of muscle activation intervals were calculated following the procedure reported in Section 2.3 considering first the most recurrent (Fig. 5) and then the following less recurrent modalities of muscle activation for each subject (Fig. 6). Activation intervals are represented by horizontal bars whose area is shadowed by a three-level grey scale; darker shadowing corresponds to increasing amplitude of the mean signal, according to the definition of low-level (light grey), medium-level (dark grey), and high-level (black) activation given in Section 2.3. The most recurrent pattern of activation consists of two activations for VM (observed in 61 ± 7% of total strides) and VL (62 ± 6%), from heel strike to mid-stance and from terminal swing to the following initial contact; high-level activity (black area in Fig. 5) was measured at the beginning of gait cycle (0–4% for VM and 0–7% for VL) and in the terminal swing (90–100% for both VM and VL). For RF, in the reference position RF2, the most recurrent pattern of activation during gait cycle consists of three activations (observed in 53 ± 6% of total strides), at the beginning of gait cycle,
Fig. 4. Mean (+SE) activation intervals vs. percentage of gait cycle for five different modalities with 1, 2, 3, 4 and 5 activations, respectively, detected during the walk in rectus femoris in reference position (RF2). Mean percentage frequency of each modality of activation is reported on the right-hand side of the plot. Mean and standard error refer to the whole population. The heel contact, flat foot contact, push off and swing phases are delimited by dashed vertical lines.
59
F. Di Nardo, S. Fioretti / Journal of Electromyography and Kinesiology 23 (2013) 56–61 Table 1 Activation intervals of considered muscles in their most recurrent modality of activation. Muscle
VM VL RF1 RF2 RF3 MH
First activation (% gait cycle)
Second activation (% gait cycle)
Third activation (% gait cycle)
ON
OFF
ON
OFF
ON
OFF
0.16 ± 0.09 0.13 ± 0.11 0.46 ± 0.26 0.19 ± 0.01 0.34 ± 0.21 1.32 ± 0.47
16.6 ± 2.3 15.1 ± 2.7 14.7 ± 1.4 14.4 ± 1.1 15.0 ± 1.4 10.8 ± 1.0
59.9 ± 1.2 59.5 ± 1.0 59.3 ± 1.1 45.6 ± 2.3
85.8 ± 0.8 85.5 ± 0.7 87.8 ± 0.6 87.5 ± 0.6 86.8 ± 0.5 73.9 ± 0.8
99.5 ± 0.3 99.7 ± 0.2 99.7 ± 0.2 99.8 ± 0.1 99.9 ± 0.1 99.7 ± 0.1
46.3 ± 1.7 47.0 ± 0.8 45.9 ± 1.2 35.0 ± 2.0
Activation intervals are expressed as the timing, in the percentage of gait cycle (GC), of signal onset and offset for vastus medialis (VM), vastus lateralis (VL), medial hamstrings (MH), and rectus femoris in position 1 (RF1), position 2 (RF2) and position 3 (RF3). Values are expressed as mean ± inter-subject SE.
Fig. 5. Quantized average patterns of muscle activation intervals vs. percentage of gait cycle (GC) for vastus medialis (VM), vastus lateralis (VL), medial hamstrings (MH), and rectus femoris in position 1 (RF1), position 2 (RF2) and position 3 (RF3), in their most recurrent activation modality. Mean activation intervals are represented by horizontal bars using a three-level grey scale; darker shadowing corresponds to increasing amplitude of the mean signal, i.e. low-level (light grey), medium-level (dark grey), and highlevel (black) activation. The heel contact, flat foot contact, push off and swing phases are delimited by dashed vertical lines. All average data refer to the whole tested population.
around foot-off and in the terminal swing (Fig. 4, Table 1). The second most recurrent modality of activation for RF2 is similar to the 3-activation modality but with no activation in the transition between stance and swing. This 2-activation modality was observed in 26 ± 6% of total strides. In a further 14 ± 2% of total strides, four
activations were observed: at the beginning of gait cycle, around heel-off, around foot-off and in the terminal swing. The 5-activation modality, where three activations were observed around stance-to swing-transition, was relative to 3 ± 1% of total strides. The remaining 4 ± 2% of total strides was defined by a single
Fig. 6. Quantized average patterns of muscle activation intervals vs. percentage of gait cycle (GC) in the percentage of total strides (14 ± 2%) where four activation intervals were detected for rectus femoris in reference position (RF2). Mean activation intervals are represented by horizontal bars using a three-level grey scale; darker shadowing corresponds to increasing amplitude of the mean signal, i.e. low-level (light grey), medium-level (dark grey), and high-level (black) activation. The heel contact, flat foot contact, push off and swing phases are delimited by dashed vertical lines. All average data refer to the whole tested population.
60
F. Di Nardo, S. Fioretti / Journal of Electromyography and Kinesiology 23 (2013) 56–61
activation around foot-off (Fig. 4). Despite no significant differences in the total length of activation interval, in 3- (Fig. 5), 4(Fig. 6) and 5-activation modalities, the RF high-level activity (black area) measured at the beginning of gait cycle and in the terminal swing, increased approaching the vastii from RF1 to RF2, and then from RF2 to RF3. On the other hand, no significant change (or, in case, a mild decrease) in amplitude levels of RF activity around foot-off and/or around heel-off has been detected from RF1 to RF2 and then RF3. Comparable results were found for the 2-activation modality, in the transition between swing and stance.
4. Discussion In the present study, EMG signals were recorded from quadriceps femoris muscle group during gait in ten healthy adult subjects, in order to determine the picture of the activation modalities of rectus femoris (RF), based on a statistical analysis of a large number (approximately 350) of strides for each subject. The muscle activation intervals, detected in each subject at self-selected walking speed, followed roughly the typical pattern reported for each muscle during gait. However, the statistical analysis put into evidence that QF muscles showed different modalities in the number of activations and in the timing of signal onset and offset (Figs. 3 and 4). This suggests that it is worth considering not only the activation patterns of each muscle, but also how frequently they are observed. On average, the most recurrent activation pattern of the vastii was observed in a high percentage of the strides of the left lower limb; 61 ± 7% for VM and 62 ± 6% for VL (Fig. 3). It consists of two activations, from heel strike to mid-stance and from terminal swing to the following initial contact (Table 1, Fig. 5). The 2-activation modality found for the vastii is expected (Perry, 1992; Sutherland, 2001; Winter, 1991) and is required to generate tension in terminal swing in preparation for weight bearing at initial contact and to control knee flexion during weight acceptance. The contemporary recruitment of VM and VL found in this study, is consistent with previous results in healthy subjects (Cowan et al., 2001; Karst and Willett, 1995; Powers et al., 1996). The most recurrent modality of activation for rectus femoris in reference position, RF2, (observed in 53 ± 6% of total strides) consists of three activations, at the beginning of gait cycle, around stance-to-swing transition and in the terminal swing (Table 1, Fig. 4). In a further percentage (17 ± 2%) of strides, the pattern is similar, except for the activation around stance-to-swing transition that splits into two (or three) small activations, between heel-off and foot-off. In the remaining 26 ± 6% of total strides, the activity at the transient between stance and swing lacks, though activity in the swing-to-stance transition is still present. Despite the large variability in the modality of RF activation (Fig. 4), our statistical analysis points out the recurrence of two patterns of activation in the nearly totality of the strides and suggests that these patterns characterize completely the behavior of the rectus femoris during gait at self-selected speed. The first pattern of activity occurs around stance-to-swing transition and it can be monophasic (57 ± 5% of total strides), biphasic (14 ± 2%) or triphasic (3 ± 1%). This RF activity around the middle of gait cycle, reported by both sEMG-based (den Otter, 2004; Ounpuu et al., 1997; Sienko Thomas et al., 1996; Sutherland, 2001), and fine-wire-based studies (Barr et al., 2010; Nene et al., 2004; Perry, 1992), has been interpreted as active participation in rapid knee extension or hip flexion, or both (Conrad et al., 1986). A lack of this activation is observed in the 26 ± 6% of total strides. No RF activity in 25% of total strides during stance-to-swing transition is also reported in a recent study on school-age children (Agostini et al., 2010). This is likely due to those strides where self-selected speed was
temporary reduced, as reported by Nene et al. (2004). The second pattern of activity occurs from terminal swing to following midstance and is present in the nearly totality of the strides (96 ± 6%). The presence of RF activity during this portion of the gait cycle has been reported by means of sEMG analysis (den Otter, 2004; Ounpuu et al., 1997; Sienko Thomas et al., 1996). The activity during early stance phase has been related to RF eccentric action to assist the vastii to control weight bearing (Shiavi, 1985; Shiavi et al., 1987; Murray, 1984); the activity in terminal swing is supposed to be present to assist knee extension and helps developing muscle tension for weight acceptance during loading response to control knee flexion (Ounpuu et al., 1997). Otherwise, Perry (1989) rejected the role of RF in participating with the vastii to provide knee extension stability during limb loading and early midstance, resting on the lack of RF from terminal swing to following mid-stance observed utilizing fine-wire EMG. This absence of signal has been reported also by more recent fine-wire studies (Barr et al., 2010; Nene et al., 2004; Perry, 1992). An explanation to this disagreement has been suggested by the observation that the apparent RF activity commonly recorded during loading response using sEMG is actually the result of cross-talk from the surrounding vastii (Nene et al., 2004; Winter et al., 1994). Annaswamy et al. (1999), however, rejected the cross-talk hypothesis, reporting three activations of RF from fine-wires data. The issue is still far from being solved. Our statistical analysis showed that, on average, the high-level activation (i.e. the peak of EMG activity) of both VM and VL occurred at approximately 0–5% and 90–100% of the gait cycle. In the 96 ± 6% of total strides considering 2-, 3-, 4- and 5activation modalities all together, the high-level activation of RF also occurred at the same relative percentage time in the gait cycle as the vastii peak (see Figs. 5 and 6 for 3- and 4-activation modality, respectively); this seems to indicate the cross-talk from the vastii as the most likely origin of this phase of the RF surface EMG. However, just because surface EMG signals recorded from different muscles are modulated similarly, it does not necessarily imply cross-talk. Thus, the present study attempted a rough evaluation of this phenomenon, by means of the analysis of the variation of the mean amplitude of RF signal detected in three consecutive positions, progressively approaching the vastii (from position RF1 to RF3, Fig. 1). The increase of RF signal at the beginning of gait cycle and in the terminal swing, and no variation from pre-swing to the beginning of the swing phase (found in the 93 ± 5% of total strides), moving from position RF1 to RF3, seems to support the cross-talk hypothesis (see Figs. 5 and 6 for 3- and 4-activation modality, respectively). However, data available in the present study do not allow to argue that RF activity measured with surface electrodes during this part of the cycle is entirely due to cross-talk activity. Thus, the possibility that RF may be active albeit at a much lower level of activity and that cross-talk simply overrates the contribution of the RF cannot be refuted. A reliable investigation on cross-talk should be performed by triggering surface EMGs in the RF muscle with the firing patterns of motor units identified, with intramuscular electrodes (Merletti and Farina, 2009), in the vasti muscles, but this is beyond the goal of the present study. Hypothesis of possible cross-talk from the hamstrings, has been excluded by the observation that the low-level MH activities in the first 10% and high-level MH activities between 80% and 90% of gait cycle do not match with high-level RF activities and no RF activities, respectively, detected in the same percentages of gait cycle (Fig. 5). Some of the results reported in the present study were achieved by interpreting quantized sEMG signals. The quantization process could cause an underestimation of the low-level EMG activity; however, this possible underestimation does not affect the study, since the reported conclusions were mostly based on the analysis of the high-level activity (signal peak) of the muscles of quadriceps group.
F. Di Nardo, S. Fioretti / Journal of Electromyography and Kinesiology 23 (2013) 56–61
5. Conclusions The processing of the surface electromyographic signal, recorded on rectus femoris (RF) over a large number (approximately 350) of strides for each tested subject, points out that RF shows different activation modalities within different strides of the same walk. The statistical analysis of these different activation modalities suggests that, during gait at self-selected speed, RF is recruited with an activation modality characterized by two patterns of activity. The first pattern, observed around stance-to-swing transition, can occur in monophasic, biphasic or triphasic shape and it is necessary to control knee extension and hip flexion from pre-swing to initial swing. The second pattern of activity occurs from terminal swing to following mid-stance in the nearly totality of the strides. The increased RF signal activity detected around the swing-tostance transition progressively approaching the vastii, suggested cross-talk from the surrounding vastii as one of the most likely origins of this phase of the RF surface EMG. However, the possibility that RF may be active albeit at a much lower level of activity and that cross-talk eventually overrate the actual contribution of the RF cannot be refuted. References Agostini V, Nascimbeni A, Gaffuri A, Imazio P, Benedetti MG, Knaflitz M. Normative EMG activation patterns of school-age children during gait. Gait Posture 2010;32(3):285–9. Annaswamy TM, Giddings CJ, Croce UD, Kerrigan DC. Rectus femoris: its role in normal gait. Arch Phys Med Rehabil 1999;80:930–4. Balestra G, Knaflitz M, Molinari F. Principles of statistical gait analysis. XIV ISEK Congress, Vienna, Austria June 22–25, 2002. p. 63–64. Barr KM, Miller AL, Chapin KB. Surface electromyography does not accurately reflect rectus femoris activity during gait: impact of speed and crouch on vasti-torectus cross-talk. Gait Posture 2010;32:363–8. Bernstein N. Coordination and regulation of movements. New York: Pergamon; 1967. Bonato P, D’Alessio T, Knaflitz M. A statistical method for the measurement of muscle activation intervals from surface myoelectric signal during gait. IEEE Trans Biomed Eng 1998;45:287–99. Conrad B, Meinck HM, Benecke R. Motor patterns in human gait: adaptation to different modes of progression. In: Bles W, Brandt W, editors. Disorders of posture and gait. London: Elsevier; 1986. p. 53–67. Cowan SM, Bennell KL, Hodges PW, Crossley KM, McConnell J. Delayed onset of electromyographic activity of vastus medialis obliquus relative to vastus lateralis in subjects with patellofemoral pain syndrome. Arch Phys Med Rehabil 2001;82(2):183–9. DeLuca CJ. The use of surface electromyography in biomechanics. J Appl Biomech 1997;13:135–63. den Otter AR. Speed related changes in muscle activity from normal to very slow speeds. Gait Posture 2004;19:270–8. Freriks B, Hermens HJ, Disselhorst-Klug C, Rau G. The recommendations for sensors and sensor placement procedures for surface electromyography. In: Hermens HJ, Freriks B, Merletti R, Stegeman D, Bok J, Rau G, et al., editors. European recommendations for surface electromyography. Results of the SENIAM project. Enschede: Roessingh Research and Development; 1999. Kadaba MP, Ramakrishnan HK, Wootten ME, Gainey J, Gorton G, Cochran GVB. Repeatability of kinematic, kinetic, and electromyographic data in normal adult gait. J Orthop Res 1989;6:849–60. Karst GM, Willett GM. Onset timing of electromyographic activity in the vastus medialis oblique and vastus lateralis muscles in subjects with and without patellofemoral pain syndrome. Phys Ther 1995;75:813–23. Merletti R, Farina D. Analysis of intramuscular electromyogram signals. Philos Transact A Math Phys Eng Sci 2009;367:357–68. Murray MP. Kinematic & EMG patterns during slow, free, and fast walking. J Orthop Res 1984;2(3):272–80.
61
Nene A, Byrne C, Hermens H. Is rectus femoris really part of the quadriceps? Assessment of the rectus femoris function during gait in able-bodied adults. Gait Posture 2004;10:1–13. Ounpuu S, DeLuca P, Bell K, Davis R. Using surface electrodes for the evaluation of the rectus femoris, vastus medialis and the vastus lateralis muscles in children with cerebral palsy. Gait Posture 1997;5:211–6. Perry J. Gait analysis - normal and pathological function. USA: Slack Inc.; 1992. Perry J, Gronley JK, Bolanger EL. Functional role of the rectus femoris in gait. In: Proceedings of the 35th Annual Meeting of the Orthopaedic Research Society, 1989, Las Vegas USA. Powers CM, Landel RF, Perry J. Timing and intensity of vastus muscle activity during functional activities in subjects with and without patellofemoral pain. Phys Ther 1996;76:946–55. Shiavi R. Electromyographic patterns in adult locomotion: a comprehensive review. J Rehabil Res Dev 1985;22(3):85–98. Shiavi R, Bugle HJ, Limbird T. Electromyographic gait assessment, part 1: adult EMG profiles and walking speed. J Rehabil Res 1987;24(2):13–23. Sienko Thomas S, Moore C, Kelp Lenane C, Norris C. Simulated gait patterns:the resulting effects on gait parameters, dynamic electromyography, joint moments, and physiological cost index. Gait Posture 1996;4:100–7. Staude G, Flachenecker C, Daumer M, Wolf W. Onset detection in surface electromyographic signals: a systematic comparison of methods. EURASIP J Appl Signal Process 2001;2:67–81. Sutherland DH. The evolution of clinical gait analysis part l: kinesiological EMG. Gait Posture 2001;14:61–70. Winter DA. The biomechanics and motor control of human gait: normal elderly and pathological. 2nd ed. Ontario: Waterloo Biomechanics; 1991. Winter DA, Fuglevand AJ, Archer SE. Cross-talk in surface electromyography: theoretical and practical estimates. J Electromyogr Kinesiol 1994:415–26. Winter DA, Yack HJ. EMG profiles during normal human walking: stride to stride and inter-subject variability. Electroencephalogr Clin Neurophysiol 1987;67:402–11.
Francesco Di Nardo received his Ph.D. in Artificial intelligence systems from Università Politecnica delle Marche, Ancona, Italy in 2005. His main research activity involves the development and the clinical application (type-2 diabetes, insulin resistance and hypertension) of mathematical models of metabolic and endocrine systems. In this field, he published several papers in peer review journals and participated in National research projects. Recently, he is addressing his research in the field of movement analysis for motor rehabilitation, with particular involvement in the acquisition and the processing of surface electromyography signal to assess the muscular function during gait task.
Sandro Fioretti graduated in 1979 in Electronic Engineering at Ancona University and presently is Associate Professor in Bioengineering at the Department of Information Engineering – Università Politecnica delle Marche – Ancona. He teaches Movement Biomechanics and Bioengineering of Motor Rehabilitation at the Biomedical Engineering course of the same University. His main research interests are in the field of human movement analysis and its related fields such as: stereophotogrammetry, linear and nonlinear filtering, joint kinematics, analysis and identification of postural control, static and perturbed posturography, gait analysis, dynamic electromyography. He participated in various European and National research projects in the field of movement analysis for motor rehabilitation. He is author of numerous scientific publications in international journals, books and congress proceedings.