tyti.'' o, Department ofVeterans Affairs

Journal of Rehabilitation Researchand Development Vol . 26 No . 3Pages 45—52

Noninvasive measurement of the stiffness of tissue in the above-knee amputation limb

Mary Malinauskas, MS ; Thomas A . Krouskop, PE, PhD ; PAy A . Barry, MSDepartment of Rehabilitation, Baylor College of Medicine and VAMC, Houston, TX 77030

Abstract—In order to characterize the mechanical proper-ties of the soft tissue in above-knee amputations, ninesubjects were measured with a Doppler ultrasound sys-tem. Measurements were made at four locations : anterior,posterior, lateral, and medial . The characterizationsincluded tissues up to 2 .5 cm deep. The average posteriormoduli are significantly greater than the anterior andlateral parameters . No significant difference was foundamong moduli from the medial zone compared to otherareas . Superficial tissue had a significantly higher modu-lus than the tissue beneath . A simple method fortransducer placement produced repeatable results . Thepresent technique proved to be useful with patients ingood health, and with no severe residual limb complica-tions . The information generated with the ultrasonicdevice may aid in prosthesis fitting and will be used in aComputer-Aided Design (CAD) system as well as in otherclinical applications.

Key words : above-knee amputation, CAD/CAM,Doppler ultrasound, prosthesis fitting, residual limbcomplications, tissue characterization.

INTRODUCTION

Prosthetists measure tissue stiffness to helpdetermine the condition of an amputated limb . Byassessing the tissue condition, they decide how to

Address all correspondence and requests for reprints to : Thomas A.Krouskop, PE, PhD, Department of Rehabilitation Engineering, BaylorCollege of Medicine, 1333 Moursund Avenue, Houston, TX 77030 .

shape the socket. Traditionally, these judgmentsave been subjective . Published guidelines aid in

assessing the amount and condition of soft tissue inthe amputated limb; however, these are qualitativedescriptions of the tissue, such as "soft" or "av-erage" (10) . These types of measures are sufficientfor an experienced prosthetist, but a new prosthetistmay not find it easy to make the required judgment.Additionally, recent developments in Computer-Aided Design and Computer-Aided Manufacturing(CAD/CAM) for prosthetic fitting have fostered theneed for the quantification of many of the parame-ters that formerly were measured intuitively by theprosthetist.

Our research in CAD has included the develop-ment of a tissue modulus measurement device . Thedevice collects data that can be used in the calcula-tion of a measurement of stiffness, the Young's orelastic modulus . The resulting numerical value canbe incorporated into computer applications . If aCAD system is not used, the device gives theprosthetist a specific value on which to base socketrectification . Both the experienced and inexperi-enced prosthetist will have an objective measure as aguide in fitting a socket.

A device that measures mechanical propertiesof the tissue could be useful in other clinicalapplications . For example, because tissue ulcerationmay begin below the skin surface, mechanicalchanges occur in the tissues before they become

45

46

Journal of Rehabilitation Research and Development Vol . 26 No . 3 Summer 1989

apparent to the patient or clinician (7) . A device thatcould measure small changes in the mechanicalproperties of the tissue would allow timely treatmentof pressure areas, promote a higher degree ofawareness in patients susceptible to ulceration, andpossibly prevent further deterioration.

Another use of such a system could be indetermining levels of muscle conditioning . If exer-cises to build muscle mass or tone were prescribed, achange in modulus of the tissue would indicatechanges in the muscle, and provide a measure of theeffectiveness of the treatment.

The equipment developed in the CAD projectmust be tested thoroughly to determine the normalranges of tissue moduli, the differences in modulifor "fatty" as opposed to "muscular" tissue, andthe differences in moduli relating to sex, age,weight, time since amp atation, cause of amputation,and other relevant variables . The purpose of thispreliminary study was to determine the presentcapabilities of the system and identify areas ofimprovement that would make the system moreuseful clinically.

Mechanical properties of tissue are of signifi-cant interest to the medical community. Manyinvestigators have studied and tested different typesof tissues, but the mechanical characteristics ofhuman tissue, especially muscle and other softtissues, are incompletely tabulated at this time . Invivo tissues are nonlinear, inhomogeneous, aniso-tropic materials which exhibit many complicatedbehaviors such as creep, stress-relaxation, and hys-teresis (2,4,9).

Additionally, most studies of human tissueshave concentrated on static or dynamic testing ofcadaver parts . Generally, these tests were conductedto the tissue's failure point . Ultimate strength valuesare useful in finding replacement materials whenneeded ; however, they provide little insight into thebehavior of the tissue during normal use in the body.

Though more difficult to measure, the elasticmodulus of the tissue is of more interest for socketdesign than is the ultimate strength of the softtissue . In this regard, some elastic modulus valueshave been presented . On a basic level, the Young'smodulus of elastin (the primary component in elasticbody tissues) is approximately 600 kPa, and that of:ollagen is 1,000 MPa (3) . Ligaments have moduli inhe 400-600 MPa range (1,11) . These values were allneasured on tissues along their long axis, i .e ., for

tendons, ligaments, and muscles, along their line ofaction; and for skin, parallel to its surface . In manyloading conditions, especially in the prosthetic fit-ting, the loads are transverse to the line of action,such as the pressures which compress : a) the legsand buttocks while sitting ; b) the bottom of the footinside the shoe while standing ; and, c) the residuallimb inside its socket while standing . Body tissuesare anisotropic : their mechanical characteristics arenot the same in all directions, and so tensilestrengths or even elastic moduli measured along thelength of the tissue provide incomplete insight intothe mechanical behavior of the tissue during differ-ent loading situations.

Moreover, much of the research reported in theliterature has been completed on isolated tissues;that is, generally muscle was tested with fat removed(4). Ligaments and tendons, though sometimes test-ed with small sections of bone, were usually sepa-rated from their surroundings . Body fluids andnormal spaces between tissues were not included . Itis very difficult to project the mechanical behaviorof the system using data collected from isolatedcomponents.

In order to measure the elastic modulus of alive tissue in its functional position, an ultrasonicmeasurement device was developed (8) . Ultrasoundis noninvasive, painless, familiar to clinicians andmany patients, and ultrasonic devices are widelyavailable.

METHODS

The modulus of elasticity was measured on theresidual limb tissue of nine adult males with above-knee amputations . All amputations were performedat least one year prior to this study . In thispreliminary study, subjects were not categorizedaccording to age, cause of amputation, or crutchusage. These variables should be investigated sepa-rately in subsequent studies.

Each subject sat in a sling apparatus over themeasurement equipment . An opening in the sling ac-commodated the subject's residual limb, which hungvertically into the transducer apparatus (Figure 1).

Moduli were measured with a pulsed Dopplerultrasonic device (8) . This device is similar to thetype of ultrasound apparatus commonly used formeasuring blood velocity (5,12) . Normally, these

47

MAL1NAUSKAS et al .

Noninvasive Measurement of Tissue Stiffness

Figure 1.Subject in sling apparatus.

instruments measure the phase of the returningsignal . When a sinusoidal acoustical waveform(pulse) is applied to the tissue, the returning wavehas changed in phase after hitting the target particle.This phase shift is measured by a quadrature phasedetector and is directly proportional to the distancefrom the transducer . As the target moves, the phaseof the corresponding signal also changes . Therefore,the movement of the target can be "tracked" bycontinuously monitoring the phase of the returningsignal.

Because the residual limb tissue is virtuallymotionless when placed in the measurement device,it is necessary to move the tissue in order to measuredisplacement . This motion, or "perturbation" ofthe tissue is accomplished by pushing on the surfaceof the skin with the head of the measurement device(see Figure 2) . This head contains the ultrasoundtransducer, which also moves during the perturba-tion . The head is curved to fit the contour of the legand is 5 cm in diameter, so that a relatively largearea, compared to the size of the transducer, is

moved for measurement . This movement of a largeamount of tissue causes behavior which can bedescribed as a "flow" of tissue, similar to the fluidflows that the Doppler normally measures . In thismanner, a nearly uniform motion of the tissuesurrounding the transducer takes place, with theexception of the extremities of the measurementfield . The outermost and innermost layers generallycompress more than does the middle section of thesoft tissues because of their juxtaposition withsignificantly harder materials (the measurementhead on the skin surface, and the bone on the innerpart of the tissue) . These sections are not consideredin the data analysis . The tissue in the measurementregion is compressed slightly to keep the tissue incontact with the transducer, but this compression isassumed to have a negligible effect on the results.

For measurement, the frequency and amplitudeof motion had to be optimized for accurate datacollection at the ultrasound probe frequency . Thetransducer frequency used was 10 MHz, with a pulserepetition frequency of 4 kHz . The perturbation

48

Journal of Rehabilitation Research and Development Vol . 26 No . 3 Summer 1989

Figure 2.

Ultrasound transducer and head of perturbation device.

frequency was set to 8 Hz and the amplitude of thehead motion was 0 .25 cm. This optimized themotion for measurement and was comfortable to thesubject because the tissue deformation was small.

An Apple lie computer was used for dataacquisition and data analysis . The peak displace-ments of the motion cycle were located by samplingthe output displacement signal every 10 millisec-onds . A running average of peaks was calculatedand randomly sampled . If the value did not vary fora total of at least five samples, the value of thedisplacement was recorded and the depth waschanged to measure the next target . If the samplesdid not match, the measurement process continueduntil five consecutive equal values were obtained.The depth resolution varied with increasing depth.The highest resolution occurred between 0 .5 and 1 .5cm, and the lowest after 2 cm. The poorestresolution was at approximately 0 .1 cm. This isconsistent with the usual behavior of 10 MHzDoppler devices (6) .

After the displacement data were acquired, theywere plotted on displacement-versus-time curves(Figures 3 and 4) . Values were drawn from thesecurves for use in the modulus equation:

U2(X3

Xi)

[1]

X2 –

– X3 – X.2

where w is the frequency of cyclic displacementof the tissue,

p is the tissue density, 1 .1 gm/cm3 ,Xi is the ith depth in the tissue,and U i is the motion amplitude at depth X i .

This equation uses three representative displace-ments measured by the Doppler system (representedby U I , U2, and U3), and their corresponding depths(X 1 , X2, and X3 ) to calculate the modulus . Themodulus also depends on the density of the tissue,which is considered to be a constant, and thefrequency of the applied motion to the tissue, 8 Hz.

E = 1 .5pw 2Ul U2 U2 U3

49

MALINAUSKAS et at .

Noninvasive Measurement of Tissue Stiffness

Equation [1] was theoretically derived from the lawsof physics, and further details concerning thisderivation, as well as the sensitivity to the assump-tions of constant density and homogeneity, havebeen published previously (6).

Because the modulus equation (Equation [1])uses only three data points from the whole data set,the modulus depends upon which points are chosen.Often, stray points occur because of movement ofthe subject or large inhomogeneities in the measure-ment field (such as scar tissue or large bloodvessels), and these points tend to change the calcu-lated modulus . These values are not correct, becausethe modulus should depend only on the materialtested, and the choice of any three points from onedata set should yield the same modulus . To mini-mize the effect of erroneous points, the data were"smoothed" to fit the exponential curve of the formy = Ce - where C and k are characteristicconstants of the particular curve, determined fromthe data . This form was chosen because an idealelastic material would produce an exponentiallydecaying curve. Using this smoothing technique, themodulus varies only slightly when different sets ofthree points are chosen, eliminating dependenceupon point choice . The modulus, then, is onlyrelated to the characteristics of the curve, howquickly or slowly it decays.

Repeatability tests were performed before anysubject testing began. For convenience, modulusreadings were taken on the largest surface of thelower leg (over the gastrocnemius muscle) of onesubject. Although these values are not comparableto data from the residual limbs of amputees, theexperimental and analytical techniques were thesame, and so the tests are valid for checking therepeatability of the measurement technique.

For this evaluation, five sets of readings weretaken without removing the apparatus . Because theequipment is intended to be used to measure thesame region of a patient on several occasionsthroughout the patient's treatment, it is necessary tobe able to replace the instrument in approximatelythe same location without marking the skin. Itwould be convenient to estimate the location ofprevious readings without referring to anatomicallandmarks . To test the repeatability of measure-ments taken in this manner, the perturbator wasremoved and then, with no markings made ormeasurements taken, replaced for five more read-

ings . The entire procedure was repeated again (i .e .,five more measurements) . One week later, five finalmeasurements were taken in the same region toinsure that the values were repeatable from day today as well.

Once repeatability was ascertained, subject test-ing was performed . Measurements at four areas ofthe residual limb were taken on each subject . Thesewere: a) an anterior position of the residual limbover the rectus femoris muscle group ; 2) a posteriorposition over the semitendinosus muscle group ; 3) amedial area where there appeared to be the leastamount of developed muscle present ; and, 4) alateral zone approximately 180 degrees from themedial measurement point . Each zone was approxi-mately halfway between the distal and proximalends of the residual limb. These areas varied slightlyon each subject due to different residual limb sizesand shapes, scar tissue, and muscle condition.However, an attempt was made to place the trans-ducer at the regions of interest as consistently aspossible.

After all data were collected, they were manu-ally compiled and analyzed for variance in modulusvalues between locations and depths . Significantdifferences were determined using standard two-tailt-tests and F-tests with the significance level set at0.01.

.25

Ec)

Ga)Ea)va:

N_

0 0

1

2

3

Depth (cFigure 3.Displacement data from Doppler ultrasound system .

50

Journal of Rehabilitation Research and Development Vol . 26 No . 3 Summer 1989

__3

Depth (cm)

Figure 4.Ultrasound data, example of division of data.

During the analysis, it was recognized that attimes the perturbator device did not contact the skinwell due to unusual contours of the residual limb,movement of the patient, or loosening of thebracket . In these cases, the data set appears unusu-ally "flat," and yields an abnormally high modulusvalue . In order to exclude these errant readings, allvalues that deviated from the mean by over 100percent were discarded . Nine such points wereeliminated from a total of 103 modulus values.

RESULTS

Resulting data were plotted on displacementversus depth graphs (see Figure 3) . These are easilycomparable for different measurements, a stiffermaterial giving a generally flatter curve, and aless-stiff material resulting in a more sharply declin-ing curve.

While examining the data, it was noticed thatmany of the curves had a relatively flat sectionfollowed by a sharp decline . This is easily seen inFigure 4, with the first section ending at approxi-mately 1 .16 cm . So that this characteristic was notsimply "smoothed out" during the data analysis, atwo-curve method of analysis was adopted . In thisprocedure, the data set is divided into two sections :

the modulus for each section is calculated, and,using the two previous moduli, a weighted modulusis determined for the entire curve . The point atwhich the curve is divided is determined via acomputer program which sets the division pointwhen the slope changes significantly . The averagedivision point was 0 .86 cm ± 0 .3 cm.

Data collection included displacements from 0to approximately 3 cm. In order to eliminate zonesof nonuniform displacement and poor resolution,data were discarded from 0 to 0 .2 cm and beyond2.5 cm . Often the data naturally ended before 2 .5(this will be discussed later) ; the average endpointfor all the data was 2 .15 cm ± 0 .33 cm.

Repeatability testingIn order to show that the results of the testing

were acceptably repeatable among trials, the fourgroups of five readings taken at separate test timeswere compared using an F-test analysis of variance.This showed that there were no significant differ-ences among any of the groups, regardless of havingremoved and replaced the equipment . These data areshown in Table 1.

Table 1.Reliability .

Group 1 Group 2 Group 3 Group 4

Trial 1 26 .25 33 .53 28 .96 28 .65

Trial 2 30 .74 23 .84 28 .23 29 .58

Trial 3 23 .48 23 .65 28 .63 30 .45

Trial 4 25 .96 22 .08 31 .88 29 .65

Trials 29 .33 18 .55 33 .42 19 .51

Average 27 .15 24 .33 30 .22 27 .57

Standard Deviation 2 .89 5 .57 2 .30 4 .55

Amputee testingDifferences in moduli at various locations were

determined and are presented in Table 2.Weighted modulus values for the anterior,

posterior, medial, and lateral areas were tested forvariance, again with a standard F-test . This proce-dure showed there were significant differences be-tween the groups . In order to determine whichgroup or groups were different, a least-significantdifference t-test for several means was performed.

51

MALINAUSKAS et at .

Noninvasive Measurement of Tissue Stiffness

Table 2.Average moduli by group.

Anterior Lateral Posterior Medial

No . of Data Pts . 7 8 8 7

Average 57 .9 53 .2 141 .4 72 .3

StandardDeviation = 31 .1 30 .5 79 .1 45 .5

1stCurve

2ndCurve

No . of Data Pts . = 34 30

Average = 117 .6 59 .1

StandardDeviation = 63 .0 74 .0

This test showed that the posterior modulus(mean value 141 .4 kPa ± 79 .1) was significantlyhigher than the anterior (57 .9 kPa ± 31 .1), and thelateral (53 .2 kPa ± 30 .5) values . Although it is alsohigher than the medial group (72.3 kPa ± 45 .5), itwas not significantly so.

Differences were determined for the first partof the curve, which corresponds to the superficialtissues, versus the second part of the curve, thetissue beneath . It was found that the first curve(average modulus 117 .6 kPa ± 63 .0) was significantlystiffer than the underlying tissue (average modulus59.1 kPa±74 .0).

DISCUSSION

RepeatabilityThe results of the repeatability tests indicate

that it is reasonable to estimate the location of thelast measurement for subsequent readings as long asthe same approximate position is used . Refinedequipment may be more sensitive to slight variationsin modulus, in which case more accurate placementof the transducer will become necessary . For thistest, however, the method was found to giveacceptable repeatability.

Amputee dataOne of the most striking features of the data

was the high variance of each group . Because in thisstudy no age, weight, patient condition, or other

demographic distinctions were made, it is reasonablethat the data would be highly varying . Despite this,the results show interesting trends . The posteriormuscles were, in general, the stiffest . Although theywere not significantly different from the medialreadings, it is believed that this is probably due tothe relatively small sample size of the test, and thatfurther testing would probably distinguish the twogroups.

The significance of this finding is difficult toassess . It is possible that the manner of walkingwhich most amputees adopt uses the posteriormuscles more than other areas . This could be useful,as prostheses change and improve, in determiningwhich devices alter the gait the most, and whichconfigurationis most acceptable. Is it desirable tohave this outcome, or should all the muscles be usedmore fully? This information could also be useful todetermine if a patient is learning to walk appropri-ately, once the desired modulus distribution hasbeen determined . Unfortunately, it is not knownwhether posterior stiffness is an acquired musclecondition, or if the posterior muscles have anormally higher modulus . There have not yet beenany measurements of the changes in muscle stiffnessof the various locations over time (i .e., as theamputee becomes used to walking in the prosthesis).If, in fact, the higher posterior moduli were ac-quired after the amputation, this would be reflectedin a study of modulus changes of the residual limbareas over time.

As for modulus variation throughout the depthof the tissue, the results show that the mostsuperficial, 0 .5 to 1 .1 cm depth, tissue is stiffer thanthe tissue beneath it . The first layers of tissue aregenerally fatty, and those beneath are muscular.This configuration should give a lower modulus tothe superficial tissue and a higher value for themuscle; the exact opposite was found . Skin has arelatively high tensile modulus, so the skin's confin-ing effect on the upper tissue levels may have causedthis unexpected result . This is an interesting effect,because normally a prosthesis will interact morewith the superficial layers of tissue, and the behaviorof the superficial layers will have a great effect onthe function and fitting of the prosthesis.

Another interesting outcome in this study wasthat although the ultrasound monitored the motionof the tissue to a depth of 2.5 cm, most of the datawere truncated before that point, regardless of the

52

Journal of Rehabilitation Research and Development Vol . 26 No . 3 Summer 1989

Butler DL, Grood ES, Noyes FR, Sodd AN: On theinterpretation of our anterior cruciate ligament data . ClinOrthop 196 :26-34, 1985.Crisp JDC : Properties of tendon and skin . InBiomechanics : Its Foundations and Objectives, Y .C.Fung, N . Perrone, M. Anliker (Eds .), Englewood Cliffs,NJ: Prentice-Hall, Inc ., 141-179, 1972.Fung YC : Biomechanics : Mechanical Properties of LivingTissues . New York : Springer-Verlag, 196-203, 1981.Fung YC : Stress-strain history relations of soft tissues insimple elongation . In Biomechanics : Its Foundations andObjectives, Y .C . Fung, N . Perrone, M . Anliker (Eds .),Englewood Cliffs, NJ: Prentice-Hall, Inc ., 181-208, 1972.Hartley CJ, Cole JS : An ultrasonic pulsed Doppler systemfor measuring blood flow in small vessels . J Appl Physiol27:626-629,1974.Johnston KW, Kassam MS: Processing Doppler signalsand analysis of peripheral arterial waveforms : Problemsand solution . In Noninvasive Diagnostic Techniques inVascular Disease, E.F . Bernstein (Ed.), St . Louis : C .V.

SUMMARY

Mosby Co ., 40-57, 1985.Krouskop TA : A synthesis of the factors that contributeto pressure sore formation . Med Hypotheses 11 :255-267,1983.Krouskop TA, Dougherty DA, Vinson FS : A pulsedDoppler ultrasonic system for making noninvasive mea-surements of the mechanical properties of soft tissue . JRehabil Res Dev 24(2) : 1-8, 1987.Peterson FJ, Lange WA, Gadd CW : Tear and tensilestrength of human skin under static and dynamic loading.Proceedings of the ASME Winter Annual Meeting, NewYork, NY, Paper 72-WA/BHF-6 484, 1972.Lower-Limb Prosthetics . New York, NY: New YorkUniversity, 1982.Sheh M, Butler DL, Stouffer DC : Mechanical andstructural properties of the human cruciate ligaments andpatellar tendon . Proceedings of the 32nd Annual ORSMeeting, New Orleans, LA, 236, 1986.Wagner FW: The diabetic foot . Orthopedics 10 :163-172,1987.

depth of tissue in the person's residual limb . REFERENCESAlthough there was much more tissue than 2 .5 cmon most patients, the data set was automaticallyshortened before this . The truncation point of eachdata set was determined by a computer algorithmwhich finds the point where the data no longer arereasonable . Each set decays to a certain point andthen begins to increase . This increase is physicallyimpossible and is believed to be an artifact of thecollection process . It may be caused by a largeinhomogeneity in the tissue, such as a boundarywhere muscles slip relative to each other, or where alarge blood vessel is located in the measurementfield . Also, in deeper tissues it is difficult to obtain aclear ultrasound signal . It may be possible that it issimply too difficult to collect good data in deepertissues using current ultrasound equipment .

2.

3.

4.

5.

6.

7.

The ultrasound technique for measurement ofYoung's modulus was found to be repeatable, and

8.the measurement and analysis techniques were re-fined. Preliminary results on nine male subjects withabove-knee amputation showed that the posteriormuscles have the highest modulus, and the superfi-cial tissues (first 0.5 to 1 .1 cm) exhibit higherstiffness than those beneath. Further testing toinclude tissue changes over time, stiffness of deepertissues, effects of demographic variables, differencesamong amputation causes and types, and variationsdue to prosthesis type or gait pattern are nowpossible. These results provide a basis for thedevelopment of an improved instrument that shouldbe clinically usable to quantify tissue status andprosthetic fit .

9.

10.

11.

12 .