Eur J Appl Physiol (1991) 62:180-188

u°.n A p p l i e d Physiology Journal of

and Occupational Physiology © Springer-Verlag1991

Static and dynamic assessment of the Biodex dynamometer Nigel A. S. Taylor 1, Ross H. Sanders 2, E. Ian Howick 2, and Stephen N. Stanley 2 Department of Human Movement Science, University of Wollongong, P.O. Box 1144, NSW 2500, Australia

2 Kinesiology, University of Orago, Dunedin, New Zealand Accepted October 2, 1990

Summary. The validity and accuracy of the Biodex dynamometer was investigated under static and dynamic conditions. Static torque and angular position output correlated well with externally derived data (r= 0.998 and r>0.999, respectively). Three subjects performed maximal voluntary knee extensions and flexions at angular velocities from 60 to 450 °. s - 1. Using linear accelerometry, high speed filming and Biodex software, data were collected for lever arm angular velocity and linear accelerations, and subject generated torque. Analysis of synchronized angular position and velocity changes revealed the dynamometer controlled angular velocity of the lever arm to within 3.5% of the preset value. Small transient velocity overshoots were apparent on reaching the set velocity. High frequency torque artefacts were observed at all test velocities, but most noticeably at the faster speeds, and were associated with lever arm accelerations accompanying directional changes, application of resistive torques by the dynamometer, and limb instability. Isokinematic torques collected from ten subjects (240, 300 and 400 °. s - l ) identified possible errors associated with reporting knee extension torques at 30 ° of flexion. As a result of tissue and padding compliance, leg extension angular velocity exceeded lever arm angular velocity over most of the range of motion, while during flexion this compliance meant that knee and lever arm angles were not always identical, particularly at the start of motion. Nevertheless, the Biodex dynamometer was found to be both a valid and an accurate research tool; however, caution must be expercised when interpreting and ascribing torques and angular velocities to the limb producing motion. Key words: Dynamometry - Validation - Torque - Angular velocity - Angular position

Offprint requests to: N. A. S. Taylor

Introduction Isokinematic (constant angular velocity) devices have been used to assess muscle function for over 60 years (Wyman 1926). In recent years dynamometers have become commercially available for clinical use. However, the capacity of some of these devices to fulfil the role of a research instrument has been questioned due to errors in controlling angular velocity and accounting for the effects of gravity on torque measurement (Winter et al. 1981; Sapega et al. 1982). The ability of a dynamometer to permit precise torque measurement, with minimal torque artefact, created by the application of resistive forces by the dynamometer during the control of lever arm angular velocity, is of particular importance to researchers. The nett torque within the dynamometer system is the algebraic sum of subject generated torque (dynamometer output), the resistive torque produced by the dynamometer and the torque due to gravity. When these torques sum to zero, the lever arm remains stationary or rotates at a constant angular velocity. Nett torques other than zero result in lever arm accelerations, and occur when the limb is accelerating up to the preset angular velocity, when the lever arm meets an accommodating resistive torque, when muscle activation results in the lever arm overshooting the pre-set angular velocity, and when the accommodating resistance of the dynamometer produces lever arm deceleration. Lever arm accelerations and decelerations mean the loss of isokinematic limb motion, invalidating the assumptions implicit within this mode of testing. The problems of angular velocity control and gravity correction have largely been overcome either by modifying existing apparatus so that limb acceleration is controlled (Gransberg and Knutsson 1983), by adopting suitable gravity correction methods (Winter et al. 1981), or through the development of new testing devices (Farrel and Richards 1986; Seger et al. 1988). One such new device is the Biodex dynamometer (Biodex Corporation, Shirley, NY, USA).

181 The angular velocity of the lever arm of the Biodex dynamometer is controlled by a gearbox driven by a servomotor. The hub of the powerhead, through which the central axle passes, contains four torque transducers. Output from the transducers passes to a power amplifier and then to the servomotor to initiate angular velocity control. Lever arm angular velocity is regulated by a feedback mechanism which compares the angular velocity o f the central axle (servomotor output voltage) with pre-determined reference voltages. Once the axle has reached the preset angular velocity, a resistive torque is applied to the axle, preventing it from exceeding the preset angular velocity. Torque is recorded throughout the full range of lever arm motion, with gravity correction being performed by the accompanying software, following calibration to determine the maximal effect of gravity on static torque. The d y n a m o m e t e r controller also prevents the lever arm from impacting against the mechanical range o f motion limits by decelerating the lever arm as it approaches these limits. The point for initiating deceleration is regulated by the experimenter using a dial mounted on the control panel (cushioning). The higher the cushion value chosen, the earlier limb deceleration commences during the range of motion, with this point being velocity dependent. A limitation of the Cybex dynamometer's control o f lever arm angular velocity has resulted in the production o f torque artefacts within the torque-time plot (Sapega et al. 1982). It was, therefore, the purpose o f this project to evaluate the accuracy with which the Biodex d y n a m o m e t e r controlled lever arm angular velocity during concentric isokinematic activity, and to ascertain whether the means of angular velocity control resulted in the production of torque artefacts. It was also o f interest to assess the validity and accuracy of torque and angular position output from the d y n a m o m e t e r and its accompanying software. Knee extension angle specific torque is generally reported at 30 ° of flexion by investigators using the Cybex d y n a m o m e t e r (Cybex, Chattanooga, Tenn., USA), since the Cybex produces torque artefacts at greater joint angles, particularly when operating at high angular velocities (Perrine and Edgerton 1978). It is possible that subjects may not be able to maintain high angular velocities at joint angles of 30 ° during maximal voluntary extensions, and torques recorded at this angle may be deceleration torques, with little physiological significance. If subjects were also unable to hold the preset isokinematic velocity during leg flexion, deceleration torques may similarly be obtained at angles around 60 ° . Another goal of this project was to investigate the range of knee flexion and extension through which normal subjects might be expected to perform isokinematic, concentric activity at a variety of angular velocities. Kinematic data obtained from devices such as the Biodex or Cybex dynamometers pertain to the lever arm being moved by the subject, and not necessarily to the limb in motion or to the activated muscles. The direct application of these data to the active muscle mass is influenced by several factors including the effects of

muscle length, changes in moment arm length, and the degree to which lever arm angular velocity and angular position reflect that of the limb being moved. The final purpose o f this study was to investigate the differences between limb motion and that of the d y n a m o m e t e r lever arm.

Methods The Biodex system output for angular position and torque was evaluated both under stationary (static) and subject-activated (dynamic) operating conditions, while angular velocity output was assessed under dynamic conditions. Under dynamic conditions subjects performed four maximal voluntary leg extensions and flexions over a range of angular velocities from 60 to 450°-s- ~, with varying levels of lever arm deceleration (cushioning: 0%, 25%, 50%, 75%, 100%). All torques were gravity corrected using Biodex software. Mechanical range of motion limits were always positioned at 90° of knee flexion and at full extension. Torque measurement. Torque output from the Biodex was studied statically with the lever arm in the horizontal position. Calibrated masses were attached at known distances from the axis of rotation of the lever arm (n= 11), in ascending, descending and random order. Three sets of trials were performed at inter-trial intervals of 10 months and 2 months ~. Results were analysed using linear regression analysis. Three subjects (mean age 31.7 years, SD 5.9) were studied during isokinematic extension and flexion of the right knee at 60, 180, 210, 300, 400 and 450°. s- 1. Two approaches were taken to evaluate the possible occurrence of torque artefacts as subjects brought the lever to the preset angular velocity. First, torques for maximal voluntary leg extensions and flexions were plotted against time and scanned for impact artefacts. Second, a piezoresistive accelerometer (Kyowa AS-500A uni-axial accelerometer, Kyowa Electronics Instruments Inc., Tokyo, Japan) was attached to the distal end of the lever arm. Accelerometry data were converted to linear accelerations. Linear acceleration and torque data were phase adjusted, using known and constant movement starting points. Torques (Biodex) and unfiltered linear accelerations were scanned for acceleration artefacts associated with the application of resistive torques by the dynamometer. Angular position measurement. Angular position of the lever arm

was determined statically over 10° increments from 90° (vertical) to 0° (horizontal), using a goniometer mounted on the lever arm. Data was evaluated using linear regression analysis. The device consisted of two transparent, 3600 protractors. One protractor was mounted on a ball bearing hub, and weighted to remain in a constant orientation. The second was fixed so that it remained oriented relative to the lever arm. Angular position was determined from the rotation of the first protractor relative to the second. Data were analysed using linear regression analysis. Dynamic trials were performed at six angular velocities (as above) at each of five different cushion settings (n = 3, as above). For these trials the lever arm was fitted with retro-reflective markers (2.5 cm diameter) to identify its path of rotation. Two-dimensional kinematic data were obtained using a high speed video camera (NAC model V-14B, NAC Inc., Minato-ku, Tokyo, Japan) with a 12-120 mm zoom lens (focal length 26 mm; Angenieux, Paris, France). Data were collected at 200 frames, s-1, with the optical axis of the video camera parallel to the axis of rotation of the Biodex lever arm. The paths of the reflective markers were digitized using the Flextrac-ExpertVision (FEV) motion analysis

1 The first set of data was collected by Dr R. N. Marshall and is used with his permission.

182 system (Motion Analysis Corporation, Santa Rosa, Calif., USA) and the angular displacement of the Biodex lever arm was calculated. The accuracy of linear measurements provided by the FEV system has been shown to be equivalent to that provided by more traditional cinematographic techniques (Smith et al. 1988), while one of us has recently validated the angular kinematics (unpublished observations). Displacement data were extracted from the FEV system and filtered using a second order Butterworth digital filter, employing an optimal frequency cut off. The cut off frequencies were always within the range from 5-7 Hz. Angular displacement was then calculated using the FEV system software. This technique suffered from the loss of accuracy at the higher frequencies but was quite acceptable for deriving average velocities.

Angular velocity measurement. Angular velocity of the lever arm was calculated by differentiating the filtered angular position data with respect to time. Phase plane plots (angular velocity against angular position) were derived for the lever arm and were compared with the time synchronized angular data obtained directly from the Biodex software. Synchronization was performed using known and constant movement starting points. Angular velocity data were subsequently edited to remove velocities that preceded the initial attainment of, and followed the final departure from, the preset angular velocity. Data from consecutive trials at the same angular velocity were concatenated and analysed using the FEV system to determine the deviation in mean angular velocity of the lever arm during these trials.

Physiological ranges for isokinematic operation. To determine the isokinematic range of knee motion, ten subjects performed maximal voluntary leg extensions and flexions at 240, 300, and 400° . s - ' using their preferred kicking leg. Five men (aged 19.8 years, SD 2.4; mass, 75.08 kg, SD 8.72) and 5 women (aged 18.5 years, SD 0.6; mass, 66.48 kg, SD 4.69) participated as volunteers. Data from these trials were obtained using Biodex software only and were derived as means from three repetitions performed at each angular velocity in both movement directions. Data for the male and female groups were compared separately for flexion and extension at each angular velocity using a Student's t-test.

Kinematics of the lower leg during maximal extension/flexion. To investigate the relationship between limb and lever arm kinematics, reflective markers (2.5 cm diameter) were attached at two points on the lever arm, to the right medial malleolus and to the medial condyle of the right femur. The left leg was secured in a partially abducted position to prevent accidental concealment of the markers during limb motion. High speed filming (FEV system) was used to collect angular position and velocity data for both system components from three subjects during maximal voluntary leg extensions and flexions at angular velocities from 60 to 450°. s - ' .

Results and discussion

180 160 y = 2.201+ 0.966X w / 140 = ~

120 100

~o

0

30 60 90 120 150 180 Static t o r q u e applied (Nm)

Fig. 1. Validation of static torque outpout of the Biodex dynamometer, using inert masses attached to the lever arm

Angular position and angular velocity measurement Static angular position output from the Biodex software correlated with that derived using goniometry (Fig. 2, r > 0.999). Output from dynamic trials, comprising time synchronized angular velocity and angular position data, was collected using the Biodex and highspeed filming (FEV) systems. Data collected without lever arm cushioning were superimposed to permit assessment of the accuracy with which the Biodex recorded and controlled isokinematic angular velocity and position (Fig. 3). At each of the velocities studied, it was found that the powerhead provided fine velocity control at the preset limit, with minor velocity overshoots at the point of reaching that limit. Velocity overshoots corresponded with points where the torque developed by the subject exceeded the resistive torque produced by the dynamometer. Nett torque at these points was positive during flexion and resulted in lever arm acceleration. The precision of the lever arm angular velocity control was generally greater at the higher velocities (Fig. 3), where less velocity overshoot and 'hunting' was observed. Though this trend may have been influenced by the optimal filtering technique used during data analysis, where higher frequency velocity oscillations may have been removed at the faster velocities. In general, the angular velocity and position output from the Biodex software closely followed that obtained using the FEV system. However, it appeared from the angular velocity data that while the dynamometer responded

C 100

Torque measurement 0

Torque during the static validation trials was remarkably consistent over the three trials, so all validation data were analysed together. Static torque closely approximated that applied to the horizontal lever arm, with the slope of the relationship approaching unity (Fig. 1, r = 0.998).

8o

80 2

®

M

y = -4.01 + 1.034X

8O 60 40 2O 0

r

0

0

0.9998

~.

J

-

I

I

I

I

20

40

60

80

iO0

Static angle (o)

Fig.2. Validation of static angular

position output from the Biodex dynamometer. A comparison with data obtained using goniometry. Data points are plotted as mean and SD

183 80

.... F E V

-

-

Biodex

60

40 20

160°.s -I

0 -20 -40

Start

t. --

-60

-80

_.

./

I ' I ' I ' I ' I '

-"" 70~. ,~ ~.,

z5o 200 150 100 50 o -50 -lOO ..~ -15o -?.,00 <~ -250 500 300 ~00 100 0 -100 -200 -300 -400 -500

I

I

I

I

I

I

.

. . . . . . . . . . . . . . . j f I J I 0 ZO 4 0 60 80

I I00

Angular position (o) Fig. 3. Phase plane plots of angular position and angular velocity data for the Biodex lever arm. Data represent single leg extensions (negative angular velocities) and flexions (positive angular velocities), completed in a single cycle, at angular velocities of 60, 210, and 450° .s -] (without lever arm cushioning) using Biodex software and a high speed filming system [Flextrac-ExpertVision, Motion Analysis Corporation (FEV)]. All movements are displayed in a clockwisepattern commencing at the 90° of knee flexion with sensitivity to the force applied by the subject, the Biodex software output did not reflect this control precision, since minor velocity oscillations were not presented to the observer once the lever arm attained approximately ___5% of the preset velocity (Fig. 3). This was best observed at 60 °. s-1, where a flat velocity profile was noted. One can only assume that this reflected a limitation in the software, since the manufacturers claim angular velocity control precision only up to the 5% level. Once this threshold was reached the reported angular velocity was always equal to that set by the experimenter and remained at this value until the angular velocity fell below the 5% threshold. The velocity overshoot at the point of reaching the preset angular velocity was further investigated. Though most apparent at 60 ° . s - ] (Fig. 3), it was also observed at the other velocities. The average angular velocities for four maximal leg extensions and flexions, at six preset velocities and five cushion settings, were determined using the FEV system. Five cushion levels

were used since it was unknown how the lever arm deceleration affected the velocity overshoot, or the average angular velocity attained. Average angular velocities, along with maxima, minima and standard deviations, are presented in Table 1. When analysed considering all angular velocities and cushion settings, the mean angular velocities attained during leg extensions and flexions consistently matched those set by the experimenter (r> 0.999, slope = 1.017). The correlation of this relationship was not influenced by cushion settings, providing the cushion setting did not prevent the subject from attaining the desired angular velocity. In all trials, angular velocity overshoots were observed (Fig. 3), as indicated by differences between the preset and the maximal angular velocity (Table 1). In the extreme case this difference was 14.7% of the set velocity ( - 60 °. s - ], SD 1.83°. s - 1). This error was well below that reported for the Cybex dynamometer (Sapega et al. 1982). The resultant mean angular velocity was - 6 0 . 7 8 ° - s -1 and represented a 1.3% deviation from the preset limit. The magnitude of these deviations varied between subjects but the data did provide an indication of expected differences when using the Biodex. From these data, we concluded that the Biodex could control the average lever arm velocity to within about 3.5% of the preset angular velocity (Table 1), which exceeded the 5% accuracy claimed by the manufacturers. The cushion setting was found to produce a marked effect upon both the attainment of and the period over which the subjects were able to maintain the isokinematic velocities. At all velocities the profiles of the phase plane plots were altered with increments in cushion settings (Fig. 4). This relationship was most apparent at the faster test velocities. At 400 and 450 °. s - i , a cushion setting of 75% or higher prevented subjects from reaching the preset angular velocity (Table 1, Fig. 4). It was found that a 50% cushion permitted adequate isokinematic dwell times for velocities of 300 °. s - 1 and below, while still maintaining comfortable deceleration of the lever arm. It is recommended that during the use of test velocities greater than 300 ° .s -~, the cushion be set as close as possible to the zero setting. Since cushioning did not produce a consistent trend in the angular velocities achieved, or its deviation (Table 1), one may conclude that a hard cushion should be used to ensure that physiological rather than mechanical limitations determine the capacity of a subject to move body segments at prescribed angular velocities. Synchronized time plots for maximal leg extension and flexion data obtained from the Biodex, the FEV system and the accelerometer are presented in Fig. 5. Accelerations other than zero, during the isokinematic range of motion, were produced when the resistive torque of the dynamometer failed to balance the torque applied by the subject, resulting in angular velocity fluctuations. From the current data we were not always able to identify the precise reason for individual acceleration spikes, however possible causal mechanisms may be identified. At 60 ° - s - ] large acceleration spikes were observed at the commencement of each movement phase (Fig. 5).

184 Table 1. Angular velocity parameters for the Biodex lever arm during maximal leg extension and flexion trials. Data obtained independently using a high speed filming system (Flextrac-ExpertVision, Motion Anaylsis Corporation)

Angular velocity (degrees. s - 1)

Cushion

+ 60

0 25 50 75 100 0 25 50 75 100

-

6 0

+ 180

-

1 8 0

+210

-210

+300

-300

+400

- 400

+450 Positive velocities correspond to leg flexion. Cushion settings represent a percentage of maximal cushioning, Data were edited to include velocities from the point of first reaching, to the point of final departure from preset velocities (n, number of data points). Missing data points occurred where subjects were unable to achieve these angular velocities

-

4 5 0

(%)

Mean

-

-

59.56 59.57 59.52 59.59 59.53 60.78 60.76 60.69 60.79 60.75

Maximum

-

66.58 66.57 66.35 67.46 67.27 68.82 68.63 67.17 68.42 68.49 189.30 186.85 187.31 186.31 185.28 190.41 191.10 190.84 190.48 189.06

Minimum

n

Standard deviation

-

57.66 57.96 57.64 57.83 57.71 58.17 58.49 57.78 58.58 58.11

516 520 518 500 532 504 507 506 484 494

1.56 1,57 1,54 1.71 1.59 1,83 1.87 1,68 1,86 1.88

-

173.54 173.86 180.04 180.31 180.18 180.02 180.17 180.04 180.06 180.21

308 168 60 57 55 96 99 97 94 93

3.99 4.13 2.04 1.85 1.54 3.15 3.00 3.28 3.06 2.40

0 25 50 75 100 0 25 50 75 100

178.10 179.27 183.79 183.56 182.94 185.11 - 185.23 185.67 185.73 184.53 -

-

0 25 50 75 100 0 25 50 75 100

208.65 209.48 209.85 213.46 213.12 -215.50 -215.51 -214.88 -214.56 -215.08

215.09 216.26 214.18 215.68 215.21 -221.09 -220.80 -219.60 -218.20 -221.02

204.06 204,34 206,18 210,21 210.11 -210,31 -210.20 -210.00 -210.18 -210.04

231 176 123 48 50 77 89 88 94 92

2.99 3.28 2.47 1.62 1.64 3.20 2.97 2.80 2.45 3.43

0 25 50 75 100 0 25 50 75 100

301.56 301.65 304.01 305.57 . . -304.39 -304.76 -304.93 -307.13 -302.23

308.40 307.60 308.02 309.28 . -310.61 -311.62 -310.91 -311.89 -304.58

296.40 297.45 300.07 300.37

136 97 66 43

3.37 2.89 2.55 2.72

. . -300.06 -300.00 -300.39 -300.15 -300.01

95 91 82 63 17

3.27 3.14 2.93 3.58 1.48

0 25 50 75 100 0 25 50 75 100

406.61 408.64 408.83

411.82 415.91 414.20

400.22 401.14 400.36

69 55 37

3.53 4.22 3.80

-407.51 -409.15 -410.25

-412.93 -415.68 -417.19

-400.16 -400.12 -400.16

67 55 44

3.54 4.54 5.03

460.21 462.67 454.15

471.63 472.04 456.85

450.30 450.05 451.40

48 41 19

3.79 6.18 1.83

-459.07 -459.11 -453.29

-463.87 -466.11 -455.56

-450.38 -450.67 -450.81

48 39 15

4.06 4.49 1.59

0 25 50 75 100 0 25 50 75 100

-

-

-

185

200 ~ ~ _ _ 0 ~ 100 0

210°.s-t .~~

450°.s-t

500 250

---

0 -500

~

500

o,_. 2OO

250

100 0 0

0

0 -100

-250

~-200 I

'

I

'

I

'

I

'

I

'

I

200

-500 500 250

100

0

0 -100

-250

-200

-500 0

20

40

60

80 100

0

20

40

60

80 100

Angular position (o) Three of these four spikes contained double spikes which occurred at the intitial increase in velocity (initiation of motion), or at the point of direction change of the lever arm where an angular impulse was applied by the subject to change the direction of the lever arm. The first spike may have been the result of this angular impulse, or the effect of reaching the compliance limit of the padding and strapping used to secure the limb to the lever arm. The second would appear to have represented the application of the resistive torque by the dynamometer, when the subject approached the preset angular velocity. These acceleration spikes resulted in a torque artefact, which was most noticeable during leg flexion. The magnitude of the torque artefact was small and occurred within the first 10° of motion. In all instances smaller acceleration spikes (with the opposite sign) occurred immediately after the velocity overshoot and represented the dynamometer increasing its resistive torque to bring the lever arm to the preset velocity. At the end of motion, a further torque spike occurred (particularly during flexion) but without a simultaneous acceleration spike. This latter observation was somewhat perplexing, since the FEV system revealed slight increases in angular velocity at these points which were not detected by the accelerometer. These torques occurred over the last 10° of motion and would be excluded if angular specific torques were reported. However, it peak torques were of interest then these data may be included in subsequent analyses. The origin of these torques was uncertain and caution must be exer-

O ¢.1-

,-% I

Fig. 4. The influence of lever arm cushioning (0%, 50%, 100%) on capacity to reach and maintain preset angular velocities of 210 ° .s -~ and 450 ° .s -1. Data represent single leg extensions (positive angular velocities) and flexions (negative angular velocities) moving in a clockwise pattern. Data were obtained using a high speed filming system (Flextrac-ExpertVision, Motion Analysis Corporation)

cised when considering their inclusion in data analysis. At 450 ° •s-~, while the profiles for angular velocity were smooth, it can be seen from the linear accelerations that angular velocity was constant for only brief periods (Fig. 5). This disparity resulted from filtering out the high frequency components from the FEV system output. Velocity data from the Biodex indicated a period of constant angular velocity for both extension and flexion. It may be assumed, in the absence of the relevant technical information, that filtering within the Biodex software produced this discrepancy between actual and reported angular velocity. Looking at the torque profile in Fig. 5, attention should be drawn to the region from 300 to 1000 ms. As with the 60 ° •s-~ movements, there appear to be multiple acceleration spikes as the subjects moved from flexion to extension and from extension back to flexion. Some of these spikes were possibly the superimposition of high frequency accelerations on low frequency acceleration. The first torque spike coincided with the change in movement direction where the subject decelerated the lever arm, then applied an angular impulse to accelerate it rapidly in the opposite direction. A similar spike would have occurred if the lever arm had hit the range of motion limits; however, at the higher angular velocities subjects generally did not reach these limits in either direction. Such artefacts were easily distinguished and related primarily to the external torque required to stop manually and change lever arm direction

186 60°.S - t

80

7"-"

4500.s -~

500

40

250

m

o

-40

2 ~

-250

-80 80

t'

-500 500

I

I

40

0 0

7;

250 ~'o Biode

a

I

0

J

-40 -80 10

~

"

'

I

"

'

I

-250 '

''

I

1

-500

'

d 20

o ~

0 --5

o -10

z

'

I

'

I

'

'

I

~

-20

L

-40

I

to

~ _i°d'ex

150 lOO 50

o

L r

150 " "

100 O ~ 50 0 ~ -50

0 -50 -lOO

-100 -150 v

-150 0

2000

4000 6000 0 500 Time (milliseconds)

1000

(angular impulse). A similar torque artefact early during leg extension was not apparent along the zero baseline; however, it was observed as a transient plateau offset below the baseline. A large torque artefact during flexion followed the initial spike. By examination of the FEV trace on the same plot (Fig 5), it can be seen to have coincided with the point of first reaching the preset angular velocity. An equally large torque spike was seen at the same point during extension. These were the impacts associated with the application of a resistive torque by the dynamometer. During extension, the first such spike coincided with the end of a transitory torque plateau, where the dynamometer appeared to have modified the lever arm acceleration to produce the smooth increase in velocity seen in both velocity profiles. This pattern was also apparent at the start of flexion but without a concomitant torque plateau. Subsequent acceleration spikes were interpreted as subtle dynamometer modifications of the angular velocity and ascribed to design mechanisms to minimize the torque artefact evident within the Cybex (Sapega et al. 1982). At 450 °. s-1, we expected even slight velocity manipulations to result in torque artefact. As was evident from the synchronization of the torque and acceleration curves, torque spikes generally corresponded with ac-

1500

Fig. 5. Synchronizedtime plots for angular velocity, linear acceleration and subject applied torque obtained from a Biodex dynamometer at angular velocitiesof 60°. s- ' and 450°. s-1 (zero cushioning). Angular velocities were measured using a high speed filming system [Flextrac-ExpertVision (FEV), top] and Biodex software (second graph), linear acceleration was measured with a piezoresistive accelerometer(third graph) and torques were obtained using Biodex software (bottom). Data are from a single subject

celeration spikes. It would appear that the torque overshoot present within the Cybex was minimized, though not completely removed. Since the largest torque spike for both extension and flexion coincided with the attainment of the preset angular velocity, caution must be exercised when determining torque at high angular velocities.

Physiological ranges for isokinematic operation Figures 3 and 4 demonstrate that the capacity to reach isokinematic velocity early in the movement range, and to maintain that velocity throughout a large range of motion, was determined by physiological capacities (the ability to accelerate the limb) and mechanical constraints of the Biodex (the degree of cushioning employed). Torque is frequently reported at a constant knee angle to attempt to normalize for muscle length (Perrine and Edgerton 1978). To use this approach successfully subjects must be able to develop torque at both the set angular velocity and the reference angle. Ideally, the range of motion at the preset speed should be as large as possible and the angle for torque measurement should be optimal for torque production. To

187 assess the capacity of subjects to achieve the largest possible isokinematic motion range across the faster angular velocities (240, 300, 400 ° . s - l ) , the angles at which subjects (n = 10) first attained and last fell below the preset velocities were determined (Table 2). W o m e n produced isokinematic motion over smaller movement ranges than men (Table 2), though these differences were only significant for knee extensions at 300 and 400 °-s -1 (P<0.05). The isokinematic movement range for both genders was reduced from 64.6 to 27.0 ° , and from 63.87 to 31.16 ° for the respective knee extension and flexion velocity increments (Table 2). These isokinematic movement range decrements were approximately equal for both genders. Thus, in general, as the angular velocities increased, subjects reached the preset velocity later in the range of motion and departed from velocities earlier. The angle at departure from the preset velocity during leg extension was greater than 30 ° for some subjects at each velocity tested: eight at 240 ° .s -1, nine at 300 ° .s -1 and all subjects at 400 °. s - 1. During knee flexion all subjects commenced deceleration at some angle beyong 60 ° o f flexion for all test velocities. To satisfy the criterion that angle specific torque was derived under isokinematic conditions, it was concluded that torques should be reported at some angle greater than 30 ° for extension and less than 60 ° for flexion. Taylor et al. (1989) have shown that peak leg extension torques are achieved at a mean knee angle o f 59.5 °, SD 10.2, over angular velocities from 30 to 300 ° .s -1. All the current subjects reached the preset leg extension velocity before reaching the 60 ° position, while four subjects commenced deceleration before passing through the 60 ° position during extensions at 400 °. s - 1. Similarly, Taylor et al. (1990) found leg flexion peak torques occurring at 41.0 °, SD 12.2, over angular velocities from 30 to 400 °. s-1. This joint angle fell within the isokinematic joint movement range for the two lower leg flexion velocities; however at 400 °. s - l , three subjects had not reached speed at this angle and two w o m e n were unable to achieve the required angular velocity (Table 2). It is r e c o m m e n d e d for velocities below 400 ° • s - 1 that angle specific torque should be reported at lever arm angles o f approximately 60 ° and 40 ° for leg extension and flexion, respectively.

Table 2. Angle range over which subjects (five male (M) and five female (F)) were able to maintain constant angular velocities (Biodex dynamometer) Subject

Angular velocity (degrees. s - i)

Action

Angle entering (degrees)

Angle leaving (degrees)

M1

240

Ex F1 Ex F1 Ex F1 Ex F1 Ex F1 Ex FI Ex FI Ex FI Ex FI Ex F1 Ex F1 Ex F1 Ex F1 Ex F1 Ex F1

105.0 33.3 102.0 38.7 88.7 48.0 104.3 34.0 103.0 37.3 91.0 44.3 97.3 25.0 93.7 25.7 85.7 34.3 104.3 28.0 105.7 26.7 96.0 40.3 90.7 16.7 86.3 19.7 76.3 35.3

41.0 96.3 47.3 86.7 63.0 73.7 41.3 97.0 50.0 90.0 64.0 73.7 30.3 88.7 39.3 84.7 53.0 73.7 32.7 95.3 41.7 93.0 56.7 77.3 20.7 84.7 28.0 76.7 42.3 62.7

Ex FI Ex FI Ex F1 Ex F1 Ex F1 Ex

102.3 29.0 96.0 32.0 83.0 42.7 101.0 33.0 91.7 37.0 81.7

38.3 95.0 45.7 84.7 58.7 72.0 42.3 90.0 46.3 76.3 66.7

F1

--

--

Ex FI Ex FI Ex FI Ex F1 Ex F1 Ex Fl Ex FI Ex FI Ex F1

99.7 20.0 94.3 20.0 82.0 32.3 94.3 28.0 85.7 34.7 79.7 -98.0 21.3 95.0 26.0 78.0 38.3

29.3 88.3 35.3 81.0 47.7 64.0 37.0 84.0 45.0 72.7 62.7 -33.7 87.7 42.0 80.0 57.0 67.3

300 400 M2

300 400 M3

240 300 400

M4

240 300 400

M5

240 300 400

F1

240 300 400

F2

240 300 400

F3

240 300 400

F4

240 300

Kinematics of the lower leg The capacity of the leg to follow precisely the kinematics of the lever arm is dependent upon the rigidity with which the two components are secured. All d y n a m o m eters provide p a d d e d attachments and adjustable straps to secure the limb to the lever arm. These accessories all have some degree of compliance which results in the limb deviating from the phase plane profile of the lever arm (e.g. Fig. 6). This deviation was greatest at angular velocities which permitted subjects to achieve the highest torques. At 60°.s -1 the leg extension velocity peaked at 187.5% of the average lever arm velocity,

240

400 F5

240 300 400

Angles were determined at the point of first attaining (+ 5%) and last leaving the preset angular velocity (+5%). Missing data points occurred where physiological limitations prevented the subject from reaching the preset angular velocity; Ex, extension; FI, flexion

188 120

--

Leg

....

Lever arm

80 40

0 -40

-80 -120

"-" 7

300

~" %.

200

viation is that the angle of the knee and the lever arm were not identical throughout the full range of leg flexion, particularly at the initiation of motion. I f 60 ° and 40 ° are adpoted for angle specific leg extension and flexion torque measurements respectively, then on the basis of data presented in Fig. 6, it m a y be said that leg extension velocity should exceed the preset angular velocity, while during flexion these differences should remain minimal beyond 15 °. It is suggested that methods of securing the limb to the lever arm should be adopted that minimize compression of the soft tissue and lever arm padding during all limb movement.

100

Conclusion

0

o

0

~" - I 0 0

On the basis of the above analysis it is concluded that the Biodex d y n a m o m e t e r , while displaying minor torque artefacts at the u p p e r end of the velocity range, is a valid a n d reliable testing and research tool for the investigation of gross muscle function in man. Some degree o f caution needs to be expressed concerning the unqualified interpretation o f angle specific torques and the assumption that data from the d y n a m o m e t e r m a y be ascribed entirely to muscle function and the resultant limb motion.

-200 .~

-300

500

I

I

'

I

I

'

I

. . . . . . . . .

400 300 200 100

~

0

\

-100

I

1

~

sty/

-200 -300 -400

. .

-500

'

0

I

20

I

I

40

'

I

'

I

'

I

60

800100 Angular p o s i t i o n ( )

Fig. 6. Phase plane plots of angular position and angular velocity for the Biodex lever arm and the lower leg during knee extensions (negative angular velocities) and flexions (positive angular velocities) at angular velocities of 60, 210, and 450°. s-*. Data are presented in a clockwise pattern commencing at 90° of flexion, and were obtained from different subjects using a high speed filming system (Flextrac-ExpertVision) while flexion velocity was 195.8% greater (Fig. 6). As the preset angular velocity increased so the deviation between limb and lever arm velocity declined, until at 450 ° . s -1 the respective differences were 111.1% and 106.7%. During leg extension, the limb velocity generally exceeded p e a k lever arm velocity throughout the range o f motion. Again this was most noticeable at 60 ° • s-1 and must serve to caution against the assumption that d y n a m o m e t e r output is a faithful reflection of limb motion throughout its complete range of movement. During leg flexion, the limb m o v e d some distance before the compliance limit of the accessories was reached and the lever arm started to follow the limb (Fig. 6). This occurred as the subjects changed from extension to flexion, and resulted from the compression of lever padding, the compression of soft tissues of the leg, and from a change in alignment of the central axis of the knee from the axis o f the lever arm. This deviation in position between the phase plane plots was found to be subject and velocity specific, and was minimal during leg extension. The implication of this d e -

References

Farrell M, Richards JG (1986) Analysis of the reliability and validity of the Kinetics Communicator exercise device. Med Sci Sports Exerc 18:44-49 Gransberg L, Knutsson E (1983) Determination or dynamic muscle strength in man with acceleration controlled isokinetic movements. Acta Physiol Scand 119:317-320 Perrine J J, Edgerton VR (1978) Muscle force-velocity and powervelocity relationships under isokinetic loading. Med Sci Sports 10:159-166 Sapega AA, Nicholas JA, Sokolow D, Saraniti A (1982) The nature of torque "overshoot" in Cybex isokinetic dynamometry. Med Sci Sports Exert 14:368-375 Seger JY, Westing SH, Hanson M, Karlson E, Ekblom B (1988) A new dynamometer measuring concentric and eccentric muscle strength in accelerated, decelerated, or isokinetic movements. Eur J Appl Physiol 57:526-530 Smith SL, Dillman CJ, Rosenhoover SG (1988) Validation of an automatic video analysis system. Med Sci Sports Exerc 20:$75 Taylor NAS, Cotter JD, Stanley SN, Marshall RN (1989) Muscle torque and velocity relationships in elite power and endurance athletes. In: Gregor RJ, Zernicke RF, Whiting WC (eds) Proceedings of the XII International Congress of Biomechanics, Los Angeles, USA. Abstract no. 185, pp 370-371 Taylor NAS, Cotter JD, Gartner PW, Silver DT, Stanley SN, Marshall RN (1990) The inter-relationship of torque and velocity in leg extension of normal and elite athletes. Commonwealth and International Conference on Physical Education, Sport, Health, Dance, Recreation and Leisure. Auckland, New Zealand Winter DA, Wells RP, Orr GW (1981) Errors in the use of isokinetic dynamometers. Eur J Appl Physiol 46:397-408 Wyman J (1926) Studies on the relation of work and heat in tortoise muscle. J Physiol 6l:337-352