- Open Access
Comparison of muscle activity patterns of transfemoral amputees and control subjects during walking
© Wentink et al.; licensee BioMed Central Ltd. 2013
- Received: 12 September 2012
- Accepted: 14 June 2013
- Published: 2 August 2013
Only few studies have looked at electromyography (EMG) during prosthetic gait. Differences in EMG between normal and prosthetic gait for stance and swing phase were never separately analyzed. These differences can give valuable information if and how muscle activity changes in prosthetic gait.
In this study EMG activity during gait of the upper leg muscles of six transfemoral amputees, measured inside their own socket, was compared to that of five controls. On and off timings for stance and swing phase were determined together with the level of co-activity and inter-subject variability.
Results and conclusions
Gait phase changes in amputees mainly consisted of an increased double support phase preceding the prosthetic stance phase. For the subsequent (pre) swing phase the main differences were found in muscle activity patterns of the prosthetic limb, more muscles were active during this phase and/or with prolonged duration. The overall inter-subject variability was larger in amputees compared to controls.
- Transfemoral amputee
- Spatio-temporal data
During rehabilitation transfemoral amputees learn to adapt their gait pattern to walk with a prosthesis. Several of these adaptations are already known. During gait the stance phase of the amputated limb shortens compared to that of the intact limb. Therefore the swing phase is longer for the amputated limb. The double support phase elongates when the amputated limb becomes the stance limb and shortens when the intact limb becomes the stance limb [1, 2]. The comfortable walking speed of prosthetic walkers is also lower than in normal walking [1, 3–5]. Kinematic data shows that transfemoral amputees lack plantar flexion power (push-off) at the prosthetic side. To facilitate forward propulsion they increase the work by the hip joint at the prosthetic and intact side and the plantar flexors at the intact side [4–6]. Eccentric work at the hip of the intact side decreases with respect to normal gait . Joint power during concentric knee extension increases for the intact side, with respect to normal walking .
However, little research has been performed on electromyography (EMG) during amputee gait. EMG of residual limb muscles of TFA may give valuable information on adaptations besides those that can already be found using the kinetic and spatio-temporal data . Some studies report increased and prolonged muscle activity in amputees during gait [3, 6, 8]. Bae et al.  concluded that the co-activation of the upper leg muscles of the intact limb in amputees was larger than in controls. Hong and Mun  found that during gait the muscle activity of residual limb muscles in TFA is correlated to the socket pressure. If EMG patterns are different from that of controls this might indicate specific adaptations of amputees. Muscle activity per phase (stance and swing) can give more insight in the changes in the muscle activity patterns, how they change compared to normal walking and in the adaptations amputees make when walking with a prosthesis, besides kinematic changes.
In the current study we focus on muscle activity during the stance and swing phase of prosthetic gait. Do the muscle activity patterns of the prosthetic limb change and how do they change for the stance and swing phase, compared to normal gait? We intended to have as little interventions to the prosthesis and the subsequent walking pattern as possible. Therefore we measured EMG inside the socket, without modifications, of six amputees and compared this to data of five controls. Previous studies have shown that it is possible to measure EMG with acceptable quality inside the socket of amputees [3, 10].
From this data we determined if the timings of the muscle activity changed with respect to the different phases of gait compared to normal gait. We hypothesized that the general EMG patterns during walking are comparable to those in controls, but we expected to find differences related to specific adaptations in amputees. Three muscles at the contralateral lower leg were also measured to determine the adaptations at the intact lower leg. We determined how the inter-subject variability of amputees compares to that of controls. Spatio-temporal and kinematic data were also measured for gait phase determination and to relate the results to other studies to determine the gait phases and to compare the results to previous studies.
Overview of the details of the amputees
EMG recording was performed on eight upper leg muscles in all subjects: m. gluteus maximus (GMa), m. gluteus medius (GMe), m. tensor fasciae latae (TFL), m. rectus femorus (RF), m. vastus lateralis (VL), m. biceps femoris (BF), m. semitendinosis (ST), m. adductor magnus (Add). In amputees these were measured on the residual limb, in controls these muscles were measured at one limb, which was alternated between dominant and non-dominant limb. For amputees and controls this limb will be called the “prosthetic limb” and “mimicked prosthetic limb” respectively.
At the contralateral lower limb three more muscles were measured, the m. tibialis anterior (TA), m. gastrocnemius medialis (GaM) and the m. soleus (Sol). For amputees and controls this limb will be called the “intact limb” and the “mimicked intact limb” respectively.
Electrodes were placed according to the SENIAM standards , by an experienced physical therapist. For the amputees the locations were approximated, but EMG was checked prior to the measurements by selective contraction of the muscle . On each muscle two self adhesive electrodes (Ambu, BRS) were placed as closely together as possible. EMG measurements were performed with a 16 bipolar channel Porti-system (TMSi, Oldenzaal, the Netherlands) at a sample frequency of 2048Hz, no pre-filtering was applied.
Footswitches, placed mid-heel and under the first metatarsal head of each the foot, gave information about initial contact and initial swing. Footswitch data was registered with the Porti-system.
Kinematic data were measured (100Hz) using inertial sensors from Xsens (Xsens, Enschede, the Netherlands), with 3D accelerometers, 3D gyroscopes and 3D magnetometers. Two inertial sensors were placed at the upper and lower (mimicked) prosthetic limb. Subjects wore their own low-heeled shoes.
To synchronize EMG, footswitches and inertial sensors a synchronization pulse (sync) was given at the start and end of each measurement which was visible in all data sets.
For the experiments the subjects were asked to walk at a self selected walking speed. After data recording was started, the sync was pressed and subjects started walking. After five steps they were asked to stop, turn around, wait 2-3 seconds, press the sync and walk back; this constituted one trial. Four trials were performed in all subjects.
From the footswitch data the timings of initial contact (IC), terminal stance, initial swing and loading response of each limb were determined . Foot switches were used to extract the spatio-temporal information. Full strides were cut from the EMG and inertial sensor data, from IC to IC of the (mimicked) prosthetic limb. Strides with gait initiation or termination were excluded. All strides per subject were aligned at IC of the (mimicked) prosthetic limb.
Inertial sensor data
The inertial sensor data was expressed in the body coordinate system based on a sensor-segment calibration procedure as described by Wentink et al. . This data was subsequently low-pass filtered at 10 Hz with a second order, butterworth filter. From the calibrated inertial sensor data the knee angle, hip adduction and abduction are calculated using accelerometer and gyroscope data by the method described by Watanabe et al. .
The stance and swing phase of the (mimicked) prosthetic limb were calculated per subject and expressed as percentage of the total stride time. Using the average muscle on/off timings per subject, we subsequently calculated for which percentage of the stance or swing phase the muscles were active. These were subsequently averaged for the controls and the amputees. Differences between controls and amputees were analyzed using a Kruskal-Wallis test. The level of alpha was set at 0.05.
The inter-subject variability of the EMG data was determined using the variance ratio (VR) for each subject and muscle for the stance and the swing phase [17, 18]. The VR is the variance of the data between gait cycles normalized to the total variance, whereby 0 indicates a low variance and 1 a high variance. Differences between the controls and amputees were analyzed using the Mann-Whitney-Wilcoxon test . The standard error of the mean (SEM), was calculated using , where N is the number of subjects per group .
Total stance (M)PL
C >A p =.010
Total stance (M)IL
C <A p =.008
C <A p =.010
C >A p =.008
C <A p =.045
Rotations of knee and hip
Max Hip flexion
Max Hip extension
Max hip adduction
C >A p =.006
Max hip abduction
C >A p =.002
Max knee flexion stance
C >A p =.006
Max knee flexion swing
C >A p =.029
During the (prosthetic) stance phase, the GMa of the amputee group is active for a longer period after initial contact. All other upper leg muscles are active for a similar or shorter period. In the amputee group, some muscles, become active a second time during stance; the TFL, VL, BF, ST and Add. In the controls this second phase of activity during stance for these muscles is not seen. The first period of activity shown for the RF is probably crosstalk by the VL [20, 21]. The RF becomes active just before terminal stance in controls, but shows no activity in this phase in the amputee group.
During the stance phase of the intact limb, activity of the GaM starts around the same time in controls and amputees, Sol activity of amputees starts a little earlier. The activity of the TiA in amputees continues longer during the stance phase of the intact limb, compared to controls. No significant differences were found between the activation patterns of the stance phase between controls and amputees.
The differences in muscle activity during the swing phase are larger than for the stance phase. The GMe and GMa of the amputees become active in the second half of the swing phase, whereas in the controls they become active at the end of the swing phase. This increased duration of activation is also seen for the RF, the VL and the BF. The TFL is also active at the transition from stance to swing, and has a later “second” activity onset at the end of the swing phase. The Add is active in amputees before initial swing and at the beginning of the swing phase, which is not the case in controls.
TiA activity during the swing phase of the intact limb starts later in amputees compared to controls. No differences are seen in GaM activity during the swing phase of the intact limb, but the Sol of the amputees shows activity during the first part of the swing phase, where controls do not show this activity. No significant differences were found between the activation patterns of the swing phase of controls and amputees.
Kinematic and spatio-temporal data
The kinematic data showed that the stance phase duration of the intact limb increases and the prosthetic swing phase duration also increases in amputees. This coincides with the general concept that amputees tend to stand longer on their intact limb than on their prosthetic limb, which has also been found in other studies [1, 2]. Knee flexion during initial stance differs. Controls show a knee flexion of up to 18°, in amputees this is only 4° even though all amputees had a prosthetic knee which allows knee flexion during stance. This lack of knee flexion might indicate that amputees are not comfortable using knee flexion during initial stance of the prosthetic limb, which may be caused by a lack of trust or experience in using the MPC knee to the full potential. Hip adduction and abduction are also reduced in amputees, which was also reported by Jaegers et al. . The reduction in hip adduction has most likely only a small effect on the walking pattern, as it is only a few degrees less than in controls. During normal single limb stance a small amount of adduction is seen, to ensure that the center of mass does not have to move laterally to keep it above the supporting surface. However when amputees are in prosthetic single limb support they will not bring their COM above their support surface, but keep it more medially. This can be explained by the fact that in the frontal plane they have little opportunity to correct themselves, too much lateral motion will cause a fall. This reduces the need for adduction in stance. The reduced abduction may change the walking pattern of amputees. Hip abduction is used to “shorten” the leg to ease foot clearance during the transition from the stance to the swing phase. However TFA generally find it more difficult to perform hip abduction, which makes foot clearance more difficult. The reduction in hip abduction may create the need for more adaptations from the intact limb, for instance increased plantar flexion during single intact limb support (vaulting).
The differences found in muscle activity between prothetic users and controls are mainly present in the (pre)swing phase. Muscle activity of controls resembles that of previous studies, although muscles show activity for a longer period of time [12, 22]. This may be due to the onset detection method, but the exact methods used in the previous studies were not described. Therefore it is hard to find a clear explanation for this discrepancy. Visual comparison of raw and filtered EMG data showed comparable EMG quality between controls and amputees.
When the gait stance and swing phases are compared separately, muscles in amputees do not seem to be active for much longer than in controls. At the end of the stance phase a period of activity is seen in most of the upper leg muscles, starting around the beginning of the second double support phase. This may be the mechanism by amputees to increase socket fitting at the end of the stance phase, to prepare for lifting of the prosthesis in the swing phase . Lower leg muscles of the contralateral side show a prolonged activity during stance. This increased activity could be used to ease foot clearance, ankle plantar flexion of the intact limb is used to virtually lengthen the intact limb. The prolonged activity can also be explained by the increased push-off needed from the intact limb, to propel the body forwards, to compensate for the lack of push-off on the prosthetic side. This coincides with the kinetic data, which showed increased work at the hip and plantar flexors of the intact limb [4–6].
Some of the upper leg muscles of the amputees, the BF and the VL, remain active for almost the complete swing phase. The other muscles all become active again at the end of the swing phase to prepare for initial contact. These muscles show an earlier activity onset than in controls, which may be explained by the walking strategy of amputees. Many amputees try to fully extend the knee to ensure it is locked at the end of the swing phase which is also confirmed by the reduced knee flexion during initial stance.
Our results resemble the results presented by Jaegers et al. , as far as they can be compared. They only showed muscle activity for the complete gait cycle and no exact onset timings were calculated. They also reported activity before initial swing and found differences between subjects with an amputation in the proximal or distal half of the upper leg . In the current study all amputees were amputated at the distal half of the upper leg. In some muscles the activity is slightly longer or shorter compared to Jaegers et al. . This can be due to the separated stance and swing phases in the current study and due to different approaches in detection times.
TFA showed a different activation pattern in some phases of the gait cycle, which shows that they adapt to their new prosthetic situation. Although the results show that consistent muscle activity can be measured inside the socket of TFA, the usability for prosthetic control is questionable. Variability between the amputees is higher, although patterns within the amputees are consistent. Although muscle activity patterns can change due to the disturbed anatomy by the amputation and by use of the prothesis, training may allow TFA to learn new walking patterns which in turn may need adaptations in the muscle activity patterns to control a prosthesis .
The overall inter-subject variability of the EMG data from the amputees was significantly higher than that from controls. VRs per muscle were however for none of the muscles significantly different. Granata et al.  reported VRs in healthy adults between 0.17 and 0.27, although they can go up to 0.76 in healthy adults . The main reason for a higher VR in amputees (up to 0.64) is most likely the lower walking speed .
Many of the muscles in the upper leg of the amputees are cleaved. The electrodes were placed and tested for activity according to the SENIAM standards. However, due to the amputation some muscles may have a different location and the location of the electrodes may not have been ideal. Rotations in the socket may also affect the position of the electrodes with respect to the muscle. Poor socket fitting will affect the repeatability of the signal, this will induce more noise and the prosthetic user may show more muscle activity to properly control or fit the prosthesis. One subject complained of non-optimal socket fitting as it was too large. This subject had a higher VR. None of the subjects complained about the EMG electrodes, they did not seem to effect the socket fitting. However, this does not explain the increased VR in the lower leg muscles of the contralateral limb and the hip muscles. This might indicate that the walking pattern of amputees is less consistent than that of controls. The standard deviations within amputees for spatio-temporal and kinematic data were also larger than in controls.
The amputees were a mixed group. No inclusion criteria for type of amputation or time since amputation were added. The average age (52.3) was larger than that of the controls (23). Previous studies have shown that aging may affect the spinal cord activity, walking speed and cause a higher spread in muscle activation [22, 26]. Also three prosthetic users were only prosthetic users for 5-6 month, of whom the EMG pattern may still change over time. Two of them were the elderly subjects with vascular diseases. One of these subjects also walked with a walking aid, which may also effect the muscle activity , both subjects had higher VRs. Nevertheless, even with the large variability in the group, no large deviations were seen in the EMG patterns of these subjects.
A more homogeneous and larger group of amputees with similar prosthesis may reduce the variability between the subjects. We did not find any obvious differences between the different knees, but this may also be caused by the low number of amputees. Including amputees with a short residual limb as Jaegers et al. [1, 3] did, can be an interesting addition. Measurements were performed inside the socket of the amputees. The residual limb-socket interface may have lead to increased motion artifacts, compared to using an experimental socket with build in EMG sensors. Data were checked for these artifacts. It occurred only occasionally during initial contact or initial swing that these artifacts were not removed by filtering. Trials with motion artifacts were removed, but this still allowed at least 20 steps to be included per subject. No motion artifacts were found in controls. Although we measured EMG inside the socket with reasonable quality, we did not test the reliability and validity compared to EMG measured using an experimental socket. We placed electrodes and performed EMG measurements according to the SENIAM standards, which are based on normal anatomy. No information on actual muscle locations were available, for instance from MRI. After electrode placement muscle activity was checked using selective muscle contraction. Only occasionally electrodes needed to be replaced, for a better location with respect to the muscle belly, but never more than 2-4 cm from the original placement. Therefore normal anatomy was assumed in amputees, with respect to cross-talk. Surface EMG was used for ease of electrode placement and comfort to the patient. Intramuscular EMG may have given less cross-talk and possibly more information on specific muscle activity, but it is impossible to use in the own socket of the amputees and very uncomfortable to the patient.
In amputees the double support phase before the prosthetic stance phase increases significantly and the prosthetic swing phase shortens. EMG patterns mainly differ at the end of the stance phase and in the swing phase. These changes can explain the changes in walking strategy, but are likely also required to improve socket fitting. In this study EMG was measured inside the socket of amputees, and the data showed to be of comparable quality compared to that of controls. Variance within each amputee is higher than in controls, but variability in the kinematic data between the amputees is also higher. The increased variance may mainly be caused by the variability in walking pattern and cleavage of muscles.
This research is supported by the Dutch Technology Foundation STW, which is part of the Netherlands Organization for Scientific Research (NWO) and partly funded by the Ministry of Economic Affairs, Agriculture and Innovation, under grant no. 08003.
- Jaegers SMHJ, Arendzen JH, de Jongh HJ: Prosthetic gait of unilateral transfemoral amputees: a kinematic study. Achives Phys Med Rehabil 1995, 76: 736-743. 10.1016/S0003-9993(95)80528-1View ArticleGoogle Scholar
- Vrieling AH, van Keeken HG, Schoppen T, Hof AL, Otten B, Halbertsma JPK, Postema K: Gait adjustments in obstacle crossing, gait initiation and gait termination after a recent lower limb amputation. Clin Rehabil 2009, 23: 659-671. 10.1177/0269215509102947View ArticlePubMedGoogle Scholar
- Jaegers SMHJ, Arendzen JH, de Jongh HJ: An electromyographic study of the hip muscles of transfemoral amputees in walking. Clin Orthop Relat Res 1996, 328: 119-128.View ArticlePubMedGoogle Scholar
- Nolan L, Lees A: The functional demands on the intact limb during walking for active trans-femoral and trans-tibial amputees. Prosthet Orthot Int 2000, 24: 117-125. 10.1080/03093640008726534View ArticlePubMedGoogle Scholar
- Seroussi RE, Gitter A, Czerniecki JM, Waever K: Mechanical work adaptations of above-knee amputee ambulation. Achives Phys Med Rehabil 1996, 77: 1209-1214. 10.1016/S0003-9993(96)90151-3View ArticleGoogle Scholar
- Prinsen EC, Nederhand MJ, Rietman JS: Adaptation strategies of the lower extremities of patients with transtibial or transfemoral amputation level during level walking: a systematic review. Achives Phys Med Rehabil 2011, 92: 1311-1325. 10.1016/j.apmr.2011.01.017View ArticleGoogle Scholar
- Huang H, Kuiken TA, Lipschutz RD: A strategy for identifying locomotions modes using surface electromyography. IEEE Trans Biomed Eng 2009, 56: 65-73.PubMed CentralView ArticlePubMedGoogle Scholar
- Bae TS, Choi K, Mun MS: Level walking and stair climbing gait in above-knee amputees. J Med Eng Technol 2009,22(2):130-135.View ArticleGoogle Scholar
- Hong JH, Mun MS: Relationship between socket pressure and EMG of two muscles in trans-femoral stumps during gait. Prosthet Orthot Int 2005, 29: 59-72. 10.1080/03093640500116764View ArticlePubMedGoogle Scholar
- Huang S, Ferris DP: Muscle activation patterns during walking from transtibial amputees recorded within the residual limb-prosthetic interface. J Neuroeng Rehabil 2012,9(55):1-16.Google Scholar
- Hermens HJ, Freriks B, Merletti R, Stegeman D, Blok J, Rau G, Disselhorst-Klug C, Hagg G: European Recommendations for Surface Electromyography. Enschede: Roessingh Research and Development; 1999.Google Scholar
- Perry J: Gait Analyis; Normal and, Pathalogical Function. Thorofare: SLACK Incorporated; 1992.Google Scholar
- Wentink EC, Beijen SI, Hermens HJ, Rietman JS, Veltink PH: Intention detection of gait initiation using EMG and kinematic data. Gait and Posture 2012,37(2):223-228.View ArticlePubMedGoogle Scholar
- Watanabe T, Saito H, Koike E, Nitta K: A preliminary test of measurement of joint angles and stride length with wireless inertial sensors for wearable gait evaluation system. Comput Intell Neurosci 2011. 2011. art.ID 975193, 12 pagGoogle Scholar
- Chan ADC, Green GC: Myoelectric control development toolbox. In Proceedings 30th Conference of the Canadian Medical & Biological Engineering Society. Toronto, Canada; 2007.Google Scholar
- Zardoshti-Kermani M, Wheeler BC, Badie K, Hashemi RM: EMG feature evaluation for movement control of upper extremity prostheses. IEEE Trans Rehabil Eng 1995,3(4):324-333. 10.1109/86.481972View ArticleGoogle Scholar
- Mickelborough J, van der Linden M L, Tallis RC, Ennos AR: Muscle activity during gait initiation in normal elderly people. Gait Posture 2004, 19: 50-57. 10.1016/S0966-6362(03)00016-XView ArticlePubMedGoogle Scholar
- Hershler C, Morris M: An optimality criterion for processing electromyography (EMG) signals relating to human locomotion. IEEE Trans Biomed Eng 1978,25(5):413-420.View ArticlePubMedGoogle Scholar
- Malone A, Meldrum D, Gleeson J, Bolger C: Reliability of surface electromyography timing parameters in gait in cervical spondylotic myelopathy. J Electromyogr Kinesiol 2011, 21: 1004-1010. 10.1016/j.jelekin.2011.09.003View ArticlePubMedGoogle Scholar
- Nene A, Byrne C, Hermens H: Is rectus femoris really a part of quadriceps? Assessment of rectus femoris function during gait in able-bodied adults. Gait Posture 2004,20(1):1-13. 10.1016/S0966-6362(03)00074-2View ArticlePubMedGoogle Scholar
- Byrne CA, Lyons GM, Donnelly AE, O’Keeffe DT, Hermens HJ, Nene A: Rectus femoris surface myoelectric signal cross-talk during static contractions. J Electromyogr Kinesiol 2005, 15: 564-575. 10.1016/j.jelekin.2005.03.002View ArticlePubMedGoogle Scholar
- Rose J, Gamble JG: Human Walking. Philadelphia: Lippincott Williams and Wilkins; 2006.Google Scholar
- Granata KP, Padua DA, Abel M: Repeatability of surface EMG during gait in children. Gait Posture 2005,22(4):346-350. 10.1016/j.gaitpost.2004.11.014PubMed CentralView ArticlePubMedGoogle Scholar
- Burden AM, Trew M, Balzopoulos V: Normalisation of gait EMGs: a re-examination. J Electromyogr Kinesiol 2003, 13: 519-532. 10.1016/S1050-6411(03)00082-8View ArticlePubMedGoogle Scholar
- Merletti R, Parker P: Electromyography; Physiology, engineering and noninvasive applications. Hoboken: Wiley Interscience; 2004.View ArticleGoogle Scholar
- Monaco V, Ghionzoli A, Micera S: Age-related modifications of muscle synergies and spinal cord activity during locomotion. J Neurophysiol 2010, 104: 2092-2102. 10.1152/jn.00525.2009View ArticlePubMedGoogle Scholar
- Buurke JH, Hermens HJ, Erren-Wolters CV, Nene AV: The effect of walking aids on muscle activation patterns during walking in stroke patients. Gait Posture 2005, 22: 164-170. 10.1016/j.gaitpost.2004.09.003View ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.