7/27/2019 EMG Interpretation of Gait

1/7

Journa l o f E lec t romyography and Kines io logyVol. 1, N o. 4 . 263-269O 1991 Raven Press. L td . . New York

Technique for Interpretation of Electromyography forConcentric and Eccentric Contractions in Gait

David A . Winter and Stephen H. ScottDepartment of Kinesiology , University of Water loo, Waterloo, Ontario, Canada

Summary: We present a technique to combine muscle shortening and length-ening velocity information with electromyographic (EM G) profiles d uring gait.A biomechanical model was developed so that each muscle 's length could bereadily calculated ove r t ime as a function of angles of the joints i t crossed. Thevelocity of shortening and lengthening of the muscle fiber was then calculated,and with com puter graphics this information was overlaid on the E MG profiles.Thus, researchers and clinicians were not only able to interpret the processedEMG signal as level of activity (tension) but also to gain insight as to themuscles' role as generators (muscle shortening) or absorbers (muscle length-ening) of energy. Six common muscles are documented, using database pro-files; soleus (SO L) , medial gastrocnemius (M G), t ibialis anterior (TA) , vastusla tera l is ( VL ), rec tus femoris ( R F ) , and semitendinosus (ST). The protocolthus demonstrates a relatively simple technique for calculating muscle fibervelocity and for combining that velocity information with EMG activity pro-files. Key Words: Gait-Electromyography-Concentric-Eccentric.

Electromyograph ic ( EM G) signals in raw o r pro-cessed form are routinely used in assessment ofmuscle act ivat ion during gai t. Processed versions ofthe raw EMG signals such as the l inear envelopehave been interpreted as est imates of the relat ivetension within the muscle (7,13). These EMG pro-files provide research ers with information abou t thet ime-course of act ivi ty of the muscle but do notreflect tension-veloci ty changes and th e energy gen-erat ionlabsorption role of each m uscle over the gai tcycle. Freq uently , joint angles are measured simul-taneously with the E MG signal; this, combined withour knowledge of anatomy, offers the potential forest imating muscle length as a function of joint an -gles (2). We report an analyt ical technique for sixmajor lower l imb m uscles that al lows m uscle lengthand theref ore muscle veloci ty to be calculated fro m

Accepted April 17, 1991.Address correspondence and reprint requests to Dr. D. A.Winter at Department of Kinesiology, University of Waterloo,Waterloo, Ontario N2 L 3G1, Canada.

joint angle data. Computer graphic techniques areused to overlay this information on the EMG pro-fi les to al low more meaningful interpretat ion ofeach mu scle 's function in gai t . With this technique,researchers and cl inicians will be ab le to assess theactivity level of each muscle and simultaneouslyevalua te the role of the muscle as a genera tor(shortening veloci ty) o r as an a bsor ber (lengtheningvelocity) of energy.

THEORYResearchers and cl inical assessors can processthe raw EM G signals from ei ther surf ace or indwell-ing electrodes in many ways to obtain an analogs i gna l f rom whi ch t o i n t e rpre t movement . Thechoice of surface ver sus indwell ing electrodes is asubject of continuing argument. However, for su-perficial muscles, several advantages of surfacee lec t rodes have been demonst ra ted. The obviousease of applicat ion of surface electrodes is eviden t ,and their reliability is greatly superior ( 4 3 ) . N o

7/27/2019 EMG Interpretation of Gait

2/7

264 D . A . WINTER A N D S . H . S C O TTvalid study has yet been reported to demonstratethe superiority of either surface or indwelling elec-trodes in terms of cross-talk. Processing the EMGsignals as a linear envelope (LE) has the strongestphysiologic basis, and a rguments for suc h use havebeen presented by many researchers 1,3,7,12,14). Anumber of other processing s chemes have been in-troduced (4) without presenting any physiologic ba-sis. Figure 1 show s the similarity of a suitably cho -sen L E profi le and a muscle tension w aveform. T heL E processing involved has a transfer characterist icsimilar to that relat ing the EMG signal to muscletwi tch . The twi tch waveform is the impulse re-sponse of single motor unit (M U) in which the mo-tor unit act ion potential (M UAP ) can be consideredan impulselike neurologic input. The shape of thetwitch has been show n to have the frequency char-acterist ics of a second-order cri t ically damp ed sys -te m (6). Thus, the frequency characterist ics of acrit ically damped low-pass fil ter used in the L E pro-cessing could be chosen to mimic the twitch re-spon se. As shown in Fig. 1, second-order crit icallydamped fi l ters with cutoff of 1 . 5 4 Hz mimic twitchtimes from 106 to 40 ms, respectively. The input ofthe fi l ter is the full-wave-rectified EMG signal ,which i s a s umm at ion of al l detected M U A P s ,whe reas the output of the fi l ter mimics the summa-tion of al l twitch waveforms and therefore can beconsidered a relat ive tension waveform for thatmuscle. With suitable and somewhat complex cali-brations (7), the abso lute tension in each muscle canbe predicted, but such calibrations are too t ime-consuming and impossible to achieve (because of aloss of voluntary control) in most gait pathologies(7). Nevertheless, we are challenged to derive themost from the L E waveform as a relative tensionpredictor; we provide addit ional information re-

garding the shortening and lengthening velocity ofthe muscle.The velocities of six muscles [tibialis anterior(TA), soleus (SOL), medial gastrocnemius (MG),vastus la tera l i s (VL) , rec tus femoris (RF), andsemitendinosus (ST)] were est imated with a modelthat calculates muscle/tendon lengths using straightlines (8,lO) betw een m uscle origin and insertion (orconstraint locations, such as patel la). Th e skeletalmodel used for anatomic locations of origins, inser-tions, and co nstrain t locations relative to each jointwere derived from a scaled model for adults of thepelvis and lower limbs (10). We a ssum e that thetendons are sufficiently stiff not to lengthen appre-ciably; thus, the velocity of the origin-to-insertionlength was co nsidered equal to the muscle fiber ve-locity. The fiber velocity in meters per second wasthen divided by the resting fiber length to yield thevelocity in resting lengths per second (1,ls). Theresting length of each muscle was the length of thecontracti le component, which was assumed to beequal to i ts length when the body was in the ana-tomic posit ion. The volume of each muscle mustremain constant when i t shortens and lengthens.Th us, with pennate muscles, the angle of pennationincreases when the muscle is shortened and de-creases when it is lengthened (9).We also assumedthat active muscle and fiber lengths were propor-tional to body height. Thus, the reporting of fibervelocity in Ids was applicable for both adults andchildren.The scaled model documented by White et al .(10) was varied a t each of the hip, knee, an d anklejoints; the muscle length was calculated as a frac-tion of 1 and yielded a nonlinear relationship be-tween m uscle length and joint angles. A third-orderregression curve was adequate to re la te muscle

LINEAR ENVELOPEEMG. PROCESSING

FIG. 1. Relationship betwee n musc le tension wave-form a nd linear envelope (LE) processed electromyo-gram. The response of the critically damped low-pass filter m imics the tw itch ; thus, the low -pass fi ltercut-off f requency can be chosen to model di f ferenttwitch t imes as shown.LOW-PASS F ILTER (c r i t . damped)

J Electromyogr Kinesiol , Vol. I , No . 4 , 1991

7/27/2019 EMG Interpretation of Gait

3/7

7/27/2019 EMG Interpretation of Gait

4/7

7/27/2019 EMG Interpretation of Gait

5/7

7/27/2019 EMG Interpretation of Gait

6/7

7/27/2019 EMG Interpretation of Gait

7/7