EMG patterns during running: Intra- and inter-individual variability

.I. Electromyogr. Kinesiol. Vol. 6, No. 1, pp. 3748, 1996 Copyright 0 1996 Elsevier Science Ltd. All rights reserved Printed in Great Britain

Elsevier

10504411196 $15.00+ 0.00

ELSEVIER

1050.6411(95)ooo15-1

EMG Patterns During Running: Intra- and Inter-individual Variability Laura Guidettil , Gianfranco Rivellini2 and FrancescoFigura3 ‘Istituto Superiore di Educazione Fkica, Rome; 21stituto di Fisiologia Umana, Universitd ‘La Sapienza’, Rome; 3Dipartimento di Scienze e Tecnologie Biomediche, Universitd di L’Aquila, Italy

Summary: Rectified surface electromyographic (EMG) patterns of five healthy, young, physically-fit subjects running at 4.2 m s-l on a treadmill were recorded with the objective of defining a normal profile of EMG activity for running gait. This knowledge is important in understanding how the central nervous system (CNS) controls simple running tasks under normal conditions. The EMG signals from seven muscles (erector spinae, rectus femoris, vastus medialis, vastus lateralis, biceps femoris, tibialis anterior and gastrocnemius) were recorded, together with footswitch signals. The intra- and inter-individual variability of each muscle’s EMG profile and peak times were analysed. Interindividual EMG peak time values were analysed to define the timing of the activity of the muscles studied relative to the stride cycle and its subphases. For each muscle, little variation was found within individuals in EMG profile and peak time across trials, but differences between subjects were significant (P ~0.01). EMG peak time analysis showed two distinct activation sequences of different muscles: the first at stance phase and the second at terminal swing. In conclusion, in spite of a significant variability among subjects in EMG profile and peak time values for each muscle, the EMG peak timing analysis showed a sequence of activation at stance phase, no EMG peak activity during the first double swing and another sequence of activation during terminal swing. These findings are evidence of a neuromuscular control strategy common to all subjects. Key Words: Gait-Running-EMG. J. Electromyogr. Kinesiol., Vol. 6, 37-48, March

a normal profile of EMG activity2, similar studies on running gait are lacking. This knowledge is important in understanding how the central nervous system (CNS), under normal conditions, controls simple running tasks. In walking gait, a repeatable EMG profile for a given muscle was found within individual subjects, but not between different subjects2. The major variation depicted in that study concerned the amplitude and might reflect an intrinsic characteristic of the amplitude normalization procedure used (%MVC, i.e. amplitude as percentageof maximum voluntary contraction). The

INTRODUCTION

Electromyographic (EMG) analysis of phasic muscle activity during running has been studied by many investigators1,5,1”-15J7.However, while walking gait has been statistically analysedto define Received December 22, 1993. Revised March 31, 1995. Accepted May 15, 1995. Address correspondence and reprint requests to Laura Guidetti, Istituto di Fisiologia Umana, Universit& di Roma ‘La Sapienza’, Ple A. Moro, 5, 00185 Rome, Italy.

37

38

L. GUIDETTI

reduction of inter-subject variability by appropriate amplitude normalization has been studied for walking gait”. Some authors11,i2J5 describe the EMG activity of each muscle in terms of time of onset, while others4 describe it as time of peak activity. When the EMG signal is used to define whether the activation is on or off, there is the problem of a crosstalk signal from a neighbouring muscle*. The result of crosstalk is thus a ‘noisy baseline’, which at times may suggest a slight activity when there is none, but does not affect the quantitative interpretation at higher levels. Therefore, in spite of the methodological problem related to the on-off analysis of the EMG signal, this approach permits the evaluation of the duration of EMG muscle activation. Also, the identification of peak time of activation is not significantly affected by the crosstalk problem. Hence, EMG analysis at the instant of peak activity is well suited for investigating the neural control strategies involved in running gait. The present study was therefore designed to investigate if there is a typical pattern of EMG activity in running, through evaluating: the intraand inter-subject’s variability of EMG profile within each muscle and the inter-individual sequence of peak time activation among all muscles. METHODS Subjects Five trained, middle-distance runners, none of whom had a history of neurological or musculoskeletal injuries, participated in the study. Their anthropometric data are shown in Table 1. Prior to the experiment, the subjects signed an informed consent form.

ET AL. Experimental Protocol The subjects were allowed to exercise on the treadmill to familiarize themselves with the apparatus used to carry out the study. The running speed was 4.2 m s-l (15 km h-l) with 0% elevation, usual for middle-distance training. All subjects wore their own gym shoes, which were similar, thereby avoiding the interaction of different shoe conditions with the running patternlO. Each subject was required to perform four running tests on the treadmill. In three tests, EMG signals from two muscles of each leg were recorded. In the fourth test, the erector spinae EMG signal from both body sides was collected. Each test consisted of three to five runs, or trials, which occurred with the same speed, elevation and electrode placement conditions. Each test was followed by a 5 min rest period to avoid the onset of neuromuscular fatigue. Data Collection Electromyograms were recorded from the following muscles: erector spinae (ES), vastus lateralis (VL), vastus medialis (VM), rectus femoris (RF), biceps femoris (BF), gastrocnemius (GA) and tibialis anterior (TA). Recordings were taken from both legs; but only data from the dominant leg was then processed. The dominant leg was determined by observing which foot was placed first as the subject attempted to climb a flight of stairs, starting from a standing position with feet together. When five trials were performed, the dominant foot was used for at least three of them. Electromyograms were obtained from surface electrodes. The skin was shaved and prepared with fine sandpaper and ethanol to lower the skin impedance and favour proper recordings of the muscle potentials. Surface

TABLE 1. Anthropometric

data of subjects

Sex

Age W

Weight (kg)

Height (m)

1 2 :

M

27

M z

24 f”6

64 67 62 70

1.81 1.77 1.75 1.80

5

M

27

69

1.81

25.4 2 1.8

66.4 2 3.4

1.788 -c 0.027

Subject no.

All subjects (mean 2 sd

39

EMG PATTERNS IN RUNNING electrodes (Red Dot 3M Ag/AgCl, r = 0.3 cm, area 0.28 cm2) were placed on the belly of the investigated muscle, longitudinally to the muscle fibres, with a 2 cm centre-to-centre distance. To determine the stance and swing phases of gait cycle, both treadmill surface and subjects’ shoes were coated with conductive material so that the circuit was closed by contact with the treadmill; the stride phases were indicated by the on/off electric signal. A six-channel system was used to record data. For each leg, the electromyograms from two muscles were collected simultaneously with the footswitch signal. For each two-muscle group, between four and five steps were recorded for each of the three to five trials. The EMG signal was amplified (Figure la), bandpass filtered with a bandwidth of 50 Hz-l kHz (Figure lb) and monitored on an oscilloscope. Due to the high frequency movement artifact associated with running, it was necessary to increase the low frequency cutoff to 50 Hz. The filtered EMG signal was then full-wave rectified and low-pass filtered at 40 Hz (25 ms time constant) (Figure lc), as described in the literature6*‘. Finally, an A/D converter provided an IBM personal computer with sampled EMG signals, as well as footswitch signal inputs (sampling rate 666 Hz). Data Analysis The EMG signal was normalized to time and amplitude. The time elapsed between two heel strikes of the same leg was considered the stride time. In order to avoid the stride time variability within and between subjects, the gait-cycle time was normalized by interpolation to a fixed number of 400 samples. The amplitude was expressed as a relative value between 0 and 1, where 1 was the maximum EMG level of each gait cycle. This normalization technique was chosen to reduce the inter-individual variability of profile amplitude, as detailed in the literature’*l’. The EMG signal of each muscle was further processed in two ways: (a) amplitude values were analysed sample by sample, and (b) the timing of peaks was analysed. Mean, standard deviation and coefficient of variation (CV) were calculated for each given muscle and EMG sample; the intra-individual means, standard deviations and CV were calculated from all strides of each subject, whereas the inter-individual mean and standard deviation values were calculated from all strides of all subjects. To evaluate peak time parameters, the duration

a

b

C

LJJJ:; --I-----------0%

88%

10%

100%

TO,

d

TO,

I HS, GAIT CYCLE

FIG. 1. EMG signal collected from the biceps femoris muscle and amplified (l-10 kHz) with a HP 8811A. a, Raw EMG with movement artifacts. b, High-pass filtered (50 Hz) EMG. c, The EMG of b after rectification and smoothing. Each arrow indicates the highest point during EMG activation (EMG peak). d, Gait cycle: HS, = ipsilateral heel strike; TO, = ipsilateral toe-off; HSc = contralateral heel strike; TOc = contralateral toe-off. From HS, to TO, = ipsilateral stance phase; from TO, to HS, = ipsilateral swing phase; from TO, to HS, first double unsupported subphase (or first double swing); from TOc to HS, = second double swing (or terminal swing).

to peak was identified on the EMG profile stride by stride. We define a peak as the portion of EMG profile with a substantial amount of activity above the baseline18, indicating a higher muscle activation where the crosstalk from neighbouring muscles does not consistently affect the quantitative interpretation of activity levels. As a given EMG profile could present more than one, peaks were numbered: at stance phase no muscle showed more than one, so that peak was called ‘first peak’; during swing phase some muscles had two, so we called that which occurred initially ‘second peak’ and that which

40

L. GUIDETTl

occurred at the terminal swing ‘third peak’. An example is illustrated in Figure lc, where the BF muscle shows two peaks: one during stance phase (called first) and one at the terminal swing phase (called third): the second one (initial swing phase) is not present in this muscle. The peak time value was calculated as a percentage of the whole stride from heel strike (time zero) to instant of highest activity during the peak EMG (Figure lc). For a given muscle and peak, intraand inter-individual mean time values and standard deviation were calculated. Furthermore, we calculated, for each muscle, the number of times the EMG profile showed a given peak (frequency of occurrence) both within and among subjects. Finally, the number of times a given peak was the maximum over the entire EMG profile (frequency as maximum peak) was calculated. Statistical Analysis The intra- and inter-individual variability was quantified, sample by sample, by means of the CV. The variability of the entire EMG profile was quantified by the mean value of CV17J8. The Pearson test of normal distribution was applied, sample by sample, to test the distribution of normalized EMG data. For intra-individual EMG data, normal distribution could not be rejected. For inter-individual data, a normal distribution could not be rejected for samples in that portion of the EMG profile where the muscle was activated (peak EMG). To test significant differences between trials of each subject, the analysis of variance (one-way ANOVA) was applied to the data of the whole EMG profile; thus, intra-individual variability was tested over the whole EMG sample by sample. The inter-individual variability was assessedby ANOVA applied to portions of the EMG profile that had a normal distribution. In Figures 3a, 3b and 3c, the inter-individual mean EMG profile (marked in bold) indicates the portions over which the ANOVA was calculated. Analysis of variance was also performed (oneway ANOVA) on each ensemble peak time value to depict the presence or absence of inter-individual differences for each muscle studied. To define the sequence of activation, all muscles were first ordered by the mean time value of the considered peak, and then each peak was tested vs. the subsequent muscle by means of a Student’s t test to ascertain whether or not the muscles had a peak of activation

ET AL. at the same time. Variables of gait cycle were evaluated using means, standard deviations and ANOVA. The selected level of significance was P
Variability of EMG Profile

Intra-individual variability was quantified by means of CV values calculated sample by sample on the whole intra-individual EMG profile, as shown in Figure 2. The lower part of the figure illustrates CV values as a percentage of gait cycle, along with a mean CV (horizontal line). The mean CV value ranged from 42% (TA muscle, subject no. 5) to 66% (BF muscle, subject no. 5) (Table 3). Table 3 shows the lowest and the highest CV values (CV min and CV max) attained for a given muscle and subject. The minimum CV ranged from 19% (ES muscle, subject no. 1) to 40% (ES muscle, subject no. 4). The maximum CV value ranged from 75% (RF muscle, subject no. 3) to 122% (VM muscle, subject no. 3). A comparison between trials of the mean EMG profile of a given subject showed significant differences for only two muscles, VM and BF. Significant differences between trials were observed in portions of the EMG profile with lower activity levels, while no significant differences amongst trials were observed in portions of the waveforms at or near the peaks (Figure 2). Inter-individual

Variability of EMG Profile

Variability among subjects was quantified by means of CV values calculated sample by sample on the complete inter-individual EMG profile, as

41

EMG PATTERNSIN RUNNING TABLE 2,

Mean -t sd of stride parameters in seconds for 4.2 m s-l running speed

Parameters : 3 4 5 All subjects

0.208 0.178 0.176 0.160 0.192 0.183

0.461 0.441 0.460 0.441 0.468 0.454

kf 0.005 0.004 zt 0.002 f 0.007 f 0.008* -+ 0,017+

”t 2 2 f k

Stride

Double swing

Swing

Stance

Subject no.

0.131 0.134 0.139 0.145 0.138 0.138

0.018* 0.010 0.005 0.008 0.013’ 0.016f

f 0.012 * 0.008 k 0.004 f 0.004 * 0.006 -c 0.009’

0.669 0.620 0.636 0.601 0.661 0.637

f 0.016 zt 0.010 +- 0.005 rt 0.01 1* + 0.009 * 0.029*

Significant differences between intra- or inter-individual trials: *P
EMG variability

of each muscle

ES

VM

VL

RF

TA

GA

BF

Mean CV (%I 51 Minimum CV (%) 19 Maximum CV (% 88 Total strides 13 Trials 3 Mean CV (%) 67 Minimum CV (96) 32 Maximum CV (q/b) 78 Total strides 15 Trials 3 Mean CV (%) 49 Minimum CV (%I 20 85 Maximum CV (%) Total strides 14 Trials 3 Mean CV (%) 64 Minimum CV (%) 40 Maximum CV (%) 114 Total strides 15 Trials 3 Mean CV (%) ::, Minimum CV (%) Maximum CV (%) 77 Total strides 25 Trials 5

54 25 106 10 3 52 33 76 15 3 55 31 122 15 3

56 31 97 13 3 58 30 103 15 3 49 24 76 15 3 67 31 98 15 3 56 32 80 25 5

59 32 81 14 3 62 44 85 15 3 53 34 75 15 3 63 38 115 16 4 60 34 86 27 5

60 37 98 10 3 62 39 110 10 3 49 27 95 16 4 56 36 85 18 4 42 28 68 20 4

51 26 92 15 3 61 32 93 7 2 50 23 96 16 4 57 26 107 18 4 56 27 115 18 4

57 33 83 14 3 57 35 112 15 3 55 32 109 15 3 55 36 87 16 4 66 33 121 27 5

TABLE 3.

/r&a-individual Muscles

Subject 1

Subject 2

Subject 3

Subject 4

Subject 5

51 28 78 25 5

Mean CV, mean coefficient of variation of individual averaged EMG. Minimum CV, minimum value of coefficient of variation of individual averaged EMG. Maximum CV, maximum value of coefficient of variation of individual averaged EMG. Total strides, total individual number of strides. Trials, total individual number of trials.

shown in Figures 3a, 3b and 3c. Inter-individual mean CV ranged from 68% (VM muscle) to 90% (GA muscle). The minimum value of CV ranged from 35% (VL muscle) to 52% (TA muscle). The maximum CV valuesvaried from 100% (VM muscle) to 143% (GA muscle).

A comparison among subjects carried out by means of ANOVA on those portions of the EMG profile having normally distributed data showed statistical differences. The upper part of Figures 3a, 3b and 3c illustrate inter-individual average EMG profile and standard deviation. The portion of mean

L. GUIDETTI

l”r

BF EMG

I

HS

ET AL.

ES EMG

pm-1

*TO

HS

HS

TO

HS

cv % 113

77 43

43

% CYCLE

100

.I

50

% CYCLE

I 100

FIG. 2. Electromyographic activity (EMG) of biceps femoris muscle (BF). The graph is composed of two parts: Upper part: Intra-individual mean EMG profile and standard deviations (subject no. 5). The EMG amplitude (vertical axis) is shown as relative value between 0 and 1. Stride time (horizontal axis) is expressed as a percentage of stride cycle, with 0% and 100% corresoondina to HS: HS = heel strike. TO = toe-off. The mean EMG curve marked ‘in bold indicates’ the normally distributed portion of EMG data (point by point of the data). Significant (P ~0.01) differences among trials are indicated by means of a horizontal discontinuous line above the EMG profile. Each point on this line indicates a significant difference among trials of the corresponding points of the mean EMG curve, and the absence of points indicates no differences. Lower part: CV values referred to each sample of the intra-individual EMG profile displayed on the upper part of the graph.

FIG.3a. Electromyographic activity (EMG) of trunk muscle (ES). The graph is composed of two parts: Upper part: Inter-individual mean EMG profile and standard deviations (five subjects). For abbreviations see Figure 2 legend. The portions of mean EMG curve marked in bold indicate the normally distributed portion of EMG data (point by point of the data). Significant (P
EMG profile marked in bold represents the part of the curve where ANOVA was applied. Above the EMG profile marked in bold, a horizontal line indicates significant differences. This may be a discontinuous line due to a lack of significant differences between subjects. These differences are quantified in Table 4, as the number of samples significantly differed in percentage over the samples of the EMG profile marked in bold.

and GA presented the stance peak in almost all strides (from 98.8-100%). The terminal swing peak of activation (peak no. 3) was present in all muscles with a high frequency of occurrence ranging from 73.2-100%. Finally, the peak of activation at the initial swing phase (peak no. 2) was present in only three muscles (ES, RF, TA) (Table 5). Knee extensors and the GA showed the maximum activity level during stance phase (peak no. 1) in a high percentage over all strides ranging from 88.898.5%) whereas the TA showed the higher percentage of its maximum activity level at terminal swing (peak no. 3) as 52.8%. The BF muscle showed its maximum EMG level quite balanced between stance (peak no. 1) and terminal swing peak (peak no. 3), respectively 56 and 44%. The ES muscle presented its own maximum level of activation at initial swing peak (peak no. 2) with a percentage of 82.9% over all strides.

Variability of EMG Peak Time For each subject and muscle there were no significant differences among trials of the timing of a given peak’s EMG. Three muscles had the most consistent peak timing (Table 5). All muscles were activated during the stance phase and the frequency of occurrence of this peak of EMG activity (peak no. 1) ranged from 67.6100%. The knee extensors

43

EMG PATTERNS IN RUNNING b VM

1

EMG

--

VL

’ l-L

-

EMG

1

,!&?A.

HS

TO

HS

HS

TO

HS

cv %

cv %

I 121 --

100 67 38

$0

RF

’ l-

160

EMG

I-.-

I

J

% CYCLE

50

0

BF

l--

-

% CYCLE

EMG

-

-_

-I

100

--

p < 0.01

-

p < 0.01

r/--y

HS cv

i0

HS

I 100

.I 0

--r

TO

HS

I

1 0

FIG.3b. 3a.

HS

50

% CYCLE

50

% CYCLE

4 100

Electromyographic activity (EMG) of thigh muscles (WA, VL, RF, BF). Each graph is composed of two parts as in Figure

Inter-individual

EMG Peak Timing

All peak time values of the muscles studied are presented in Figure 4 as inter-individual mean and standard deviation as a percentage of stride. The figure shows, as dashed vertical lines, the subphases

of gait cycle. As illustrated, the muscle considered shows a peak of activation during stance and during terminal swing phases. None of the muscles studied presented an activation peak during the initial swing (first double swing). Figure 5 presents the peak time (mean and

44

L. GUIDETTI ET AL.

l;

TA

EMG

--

t

GA

‘r

--

EMG

-

HS

cv

TO

HS

HS

I

HS

CV % I

Q?I

117 -74 $-52 -.I 0

FIG.&.

50

% CYCLE

I 100

1 0

Elstiromyographic activity (EMG) of shank muscles (TA, GA). Each

TABLE 4.

Inter-individual

Muscles Peak no. 1 % Significant differences Peak no. 2 S/OSignificant differences Peak no. 3 % Significant differences

I 50

graphis amp@@

of ~WQparts aa in Figure 3a.

peak EMG variability

of each

ES

VM

Vt

RF

TA

GA

BF

100

70

95

77

100

88

92

94

98

53

94

loo

60

muscle

83 87

55

38

% CYCLE

100

% Significant differences, number (as a percentage) of EMG samples of peak EMG profile significantly different among subjects over the number of total samples of peak EMG profile (significance = P 4.01).

standard deviation) of stance phase of all muscles; statistical differences among muscles are also illustrated. Figure 6 graphically represents the peak time (mean and standard deviation) of terminal swing (second double swing). Significant differences in timing between musclesare also reported. DISCUSSION Gait Cycle Differences in the absolute lasting value in seconds of stride phases (stance, swing) were significant among different trials in three subjects; the interindividual differences were significant. Due to intraand inter-individual variability, it was necessaryto

normalize the time of each gait cyde to a fixed value. Mu&e-by-Mu&e

VarirrbWy

Intra-individual muscle EMG patterns were not different across trials, with the exception of the VM (two subjects) and BF (one subject). However, this variability was not statistically significant during the peak EMG and the CV decreased below its mean value. The significant differences observed in the portions of profile where the muscle was not activated could be due to variability caused by the crosstalk from neighbouring muscles*, In fact, the peak time value was highly repeatable among trials of a given subject, as it did not present a statistical

45

EMG PATTERNS IN RUNNING TABLE 5. Inter-individual Muscles Total strides Subjects Peak no. 1 Peak time (mean If: SD) Frequency of occurrence 1%) Frequency as peak ma% (p/o) Peak no. 2 Peak time (mean f SD) Frequency of occurrence (%I Frequency as peak max 1%) Peak no. 3 Peak time (mean -+ SD) Frequency of occurrence (%) Frequency as peak max (%I

peak EMG values of each muscle

ES 82 5

VM 65 4

VL 82 5

RF 87 5

TA 74 5

GA 74 5

BF 87 5

NS 2.1 2 1.8 92.7 13.4

5.6 i 3.3 100 98.6

* 6.3 f 3.2 100 96

6.8 i 2.8 98.8 88.8

9.5 I 4.7 67.6 16.6

* 11.1 2 5.3 100 90.6

* 15 t 6.5 95 56

* 96.4 f 3.5 74.3 9.5

NS 88.3 2 3.8 100 44

t 55 2 4 100 82.9 * 98.4 2 1.7 73.2 4

63.2 1 5.8 56.3 1 3.3 54 52.7 10.2 29.2 * 98 f 2.7 98.5 1.5

97.31 2.7 96 4

NS 98.1 + 1.6 81.8 0

* 91.1 + 4.6 91,9 52.8

*Significant peak time differences among subjects (P
difference throughout different trials. As already reported for walking gait2, the “intra-individual data obtained from a given muscle were observed to be extremely stable. This would indicate that gait might be programmed, if programming is defined as high repeatability in neuromuscular output”. Our data enablesus to infer that gait is a programmed process in running as has been suggestedfor walking. The inter-individual analysis showed statistical differences among subjects, both in EMG profile and peak timing values for each muscle. The mean inter-individual CV ranged from 68% for VM to 90% for GA muscle. These values, compared to data reported by Yang and Winter”, were lower than values referred to data either unnormalized or normalized to SO%MVC, or even normalized to isometric moment of force; but they were higher than CV values referred to EMG data normalized to peak ensemblevalue or to mean ensemblevalue. These authors reported CV values referred to the whole EMG profile to quantify inter-subject variability. However, some portions of EMG profile could vary in such a way that a CV curve calculated sample by sample would be more effective in quantifying its variability, especiallywhen the muscle was less activated. Moreover, in intra-individual comparisons the sample-by-sample CV curve was sensitive to statistical differences, rising above the

mean CV when there was a statistical significance and falling below it when there was not. On the other hand, in the inter-individual analysis, when the sample-by-sampleCV curve fell below the mean CV in correspondence to peak EMG profile, statistically significant differences could also be present. This means that there are peculiarities for individual subjects in muscular recruitment profile for a given muscle. This present study confirms, for running gait, the findings of Arsenault et al.* for walking gait, that there exist differences in the EMG profile for a given muscle among different subjects. Arsenault et al. indicated that most differences occurred in the area of the peak(s) of a given muscle profile: such differences might include the location of the peak, its amplitude and occasionallythe shapeof the profile of the envelope. They assumedthat the wide variation in amplitude depicted in their study might reflect an intrinsic characteristic due to the normalization procedure used (%MVC, i.e. amplitude as percentage of maximumvoluntary contraction). The normalization procedure utilized in the present study (i.e. the amplitude normalized to its maximum in each gait cycle), reduced the inter-individual variability”. However, we found significant differences among subjects on EMG profile. The amplitude was therefore not the major factor of inter-individual

L. GUIDETTI

46

variability. Rather, another factor, the peak location, appeared to cause differences among subjects. This was shown by the statistical analysis of the inter-individual peak time values of each muscle. The fact that intra-individual variability was low and resulted in repeatable waveform near the peaks of activity, while the inter-individual variability was consistently high, could suggest that the use of 400 points of interpolation of a single cycle of gait may be excessive and that a lower resolution could permit repeatability across subjects, For this reason, a series of different interpolation intervals were tested demonstrating that even when using 200 points of interpolation a consistently high variability across subjects in peak timing is found. Similar to findings reported for walking gait2,3, there exist individual peculiarities in the EMG pattern for a given muscle in running gait. As supported by our statistical analysis, differences in EMG profile among subjects would depend primarily on peculiar peak location for individual subjects. Sequence of Peak Activity of Studied Muscles All studied muscles presented two main peaks of activation: one at stance phase and the other at the terminal swing (double swing) phase (Figure 4).

INTER-INDIVIDUAL

ES

I I I I I I I I

*

VM

-

VL

-

RF

*

TA

-

;

GA

-

;

BF

-’ I

ET AL. During the stance phase, all muscles were activated and the peaks of activation presented the following order of occurrence: ES, VM, VL, RF, TA, GA, BF. The statistical analysis showed no differences in terms of peak timing between VM, VL and RF, indicating that these muscles were co-activated, as were GA and TA (Figure 5). As such, during stance phase we could observe an activation sequence of the muscles where the erector spinae group was the first involved, followed by the knee extensor group, then by the ankle muscles and the lastly the knee flexor. For terminal swing phase, the sequence of activation was: BF followed by TA, in turn followed by GA, VL, VM, RF and ES (Figure 6). As the foot touched the ground at the heel strike, the ES was the first muscle to present a peak of activation, its activity initiated before the instant of ground contact 16. In the first part of the stance phase, muscles controlling stabilization of both trunk (ES) and knee (VM, VL, RF) presented higher activation. In a study where electromyography and cinematography were synchronized5, the authors reported that there was a high level of VL and VM activity at initial stance phase, followed by a cocontraction of the ankle muscle to ensure stability, allowing for the downward force of the runner to be absorbed and total body stabilization achieved

INTER-INDIVIDUAL

EMG PEAK TIMING

‘LI I I I I II It I I I I I I

1 I I I

*

ES

*

VM VL

I I II 1 I I+ I I

*

RF

-

EMG PEAK TIMING

------------1 +*

-

i -

._.--.. . ,

1

TA t

GA

1 **

BF

A

SWING

% STANCE % STRIDE FIG. 4. EMG peak time values related to stride cycle for all studied muscles. Mean ? SD (horizontal line) for five subjects.

FIG. 5. Inter-individual EMG peak time mean values and standard deviation, referred only to stance phase. **Indicates significant differences (P ~0.01).

47

EMG PATTERNS IN RUNNING INTER-INDIVIDUAL

EMG PEAK TIMING

1 IES RF

VL **

i r

L

== -

GA TA

followed by the TA (dorsiflexor) and then by the other muscles. This high level of activation was often present in all muscles and represented the highest activation for TA and BF (in about half of the strides). The presence of a high activation level of muscles prior to heel strike may be attributable to the need to provide stability of ankle, leg and trunk. As suggested by Mero and KomP4, high activation levels of muscles, prior to heel strike, are needed for the increasing muscle stiffness necessary to resist great impact forces at the beginning of foot contact during running.

**

L

BF

% 2ND DOUBLE

SWING

FIG. 6. Inter-individual EMG peak time mean values and standard deviation, referred only to the second double swing phase (terminal swing phase). **Indicates significant differences (P CO.01).

in preparation for the drive phase. These authors also reported increasing EMG activity of the BF in this early support phase, with EMG activity rising to a higher level about middle stance and related to an increase in flexion of knee joint. We found that during stance a peak of knee extensors was followed first, by a peak of ankle muscle (TA, GA) activity and second, by activity of the BF. In agreement with Mann et al.‘* we observed at the initial swing phase, immediately prior to and after the toe-off (toe lifting off the treadmill), no high level of activation. During swing we observed a peak of ES muscles corresponding to heel strike of the contralateral leg, as reported by Thorstensson l6. This peak was always present and frequently (82.9%) reached the highest EMG activity value. During mid-swing the activation peak of the RF muscle could be related to leg extension, even though it was not always present (54% over all strides). As reported by Elliott and Blanksby5, the RF is not a prime mover in thigh flexion; they suggested that this function may be attributable to iliopsoas combined with a transfer of inertial force. Our data shows the presence of activation of all studied muscles at terminal swing during an unsupported phase prior to the subsequent heel strike. In this phase, the BF (thigh extensor) was the first muscle to reach a high activity level,

CONCLUSION While considering the EMG pattern muscle by muscle, we observed a highly repeatable intraindividual EMG outcome and some inter-individual peculiarities. However, pooling together results from several muscles, we were able to derive some general features concerning running gait. In fact, the inter-individual variability of peak timing for each single muscle notwithstanding, it was possible to discern a different (P ~0.01) timing of peak activation among muscles and hence define one sequence of activation during stance and one sequence during terminal swing phase. These sequences of EMG peak activation were common to all subjects. Moreover, for all subjects, there was a period of the stride during which none of the muscles studied presented a peak of activity: this relatively quiescent period occurred at the initial swing phase during the (first) double unsupported subphase. The specific activation sequence of muscles during stance, the absence of EMG peaks during the initial swing phase (first double swing) and the activation sequence of muscles during terminal swing represent the general patterns of running gait at a given velocity. Some general rules were described as EMG characters common to all subjects during walking gait2,3. Throughout all statistical analyses we noted general rules describing the ‘typical EMG pattern’ among muscles during running gait at a given velocity. These findings could conceivably be used as a template for comparison against other situations, such as in cases concerning pathology. REFERENCES 1. Adelaar RS: The practical biomechanics Sports Med 14:497-501, 1986.

of running. Am J

L. GUIDETTI

48 2. Arsenault AB, Winter DA,

Marteniuk RG: Is there a ‘normal’ profile of EMG activity in gait? Med Biol Eng

Comput 24337-343, 1986. 3. Cappozzo A, Figura F, Marchetti M, Pedotti A: The interplay

of muscular and external forces in human ambulation. J Biomech 9:35-43, 1976. 4. Dubo HIC, Peat M, Winter DA, Quanbury AO, Hobson DA, Steinke T, Reimer G: Electromyographic temporal analysis of gait: normal human locomotion. Arch Phys Med Rehabil 57:415-420, 1976. 5. Elliott BC, Blanksby BA: The synchronization of muscle activity and body segment movements during a running cycle. Med Sci Sport 11:322-327, 1979. 6. Figura F, Felici F, Macellari V: Human locomotor adjustments during perturbed walking. Hum Move Sci 5:313-332, 1986. 7. Halbertsma JM, De Boer RR: On the processing of electromyograms for computer analysis. J Biomech

14:431&S; 1981. 8. Hof AL: EMG and muscle force: an introduction. Hum Move Sci 33119-153. 1984. 9. Knutsson E, Richa& C: Different types of disturbed motor control in gait of hemiparetic patients. Brain 102:405-430, 1979. 10. Komi PV, Gollhofer A, Schmidtbleicher D, Frick U: Interaction between man and shoe in running: considerations for a more comprehensive measurement approach. Int J Sports Med 8:19&202, 1987.

ET AL. 11. Mann R, Hagy J: Biomechanics of walking, running, and sprinting. Am J Sports Med 8:345-349, 1980. 12. Mann R, Moran GT, Dougherty SE: Comparative electromyography of the lower extremity in jogging, running, and sprinting. Am J Sports Med 14:501-510, 1986. 13. Mero A, Komi PV: Force-, EMG-, and elasticity-velocity relationships at submaximal, maximal and supramaximal running speeds in sprinters. Eur J Appl Physiol55:553-561, 1986. 14. Mero A, Komi PV: Electromyographic activity in sprinting at speeds ranging from sub-maximal to supra-maximal. Med Sci Sports Ex 18:26&274, 1987. 15. Schwab GH, Moynes DR, Jobe FW, Perry J: Lower extremity electromyographic analysis of running gait. Clin Orthopaed Related Res 176:X6-170, 1983. 16. Thorstensson A, Carlson H, Zomlefer MR, Nilsson J: Lumbar back muscle activity in relation to trunk movements during locomotion in man. Actu Physiol Scund 116:13-20, 1982. 17. Yang JF, Winter DA: Electromyographic amplitude normalization methods: improving their sensitivity as diagnostic tools in gait analysis. Arch Phys Med Rehabil 65:517-521, 1984. 18. Winter DA: Electromyography in human gait. In: The biomechanics and motor control of human gait, University of Waterloo Press, Waterloo, pp 45-55, 1987.