- Email: [email protected]
Reliability of an optoelectric system to measure elbow kinematics
Clin. Biomech.
1993; 8: 315-321
Reliability of an optoelectric measure elbow kinematics
system to
T L Packer
MSc’,
PhD’,
U P Wyss Pm*, P A Costigan
‘Department of Anatomy and Division of Occupational Group, Queen’s University, Kingston, Ontario, Canada
Therapy, and ‘Clinical
Mechanics
Summary Motion analysis of the upper extremity during functional activities has only recently become more accessible. An optoelectric system (WAXMART) was subjected to testing using first a calibration dummy and then human subjects. The mean differences between the system calculated angles and those measured with potentiometers on a calibration dummy were less than 1.6” in all three planes. The test-retest reliability of the system when measuring elbow motion of human subjects yielded acceptable repeatability for measurement of functional activities. Calculation of the least significant difference found that minimum differences from 4 to 14” in flexion and from 10 to 19” in rotation can be detected using the reported set-up and protocol.
Relevance Understanding of the kinematic demands of activities of daily living will aid in the development of elbow joint prostheses and subsequent postoperative therapy. Evaluation of motor control and perceptual-motor abilities and treatment of pathological conditions will thus ensue. The first step in this understanding is knowledge of the limits of the measurement tool. This paper sets out these limits. Key words: Upper extremity, motion analysis, optoelectric, activities of daily living, elbow joint, three-dimensional
Introduction
development of technology and techniques has improved in recent years so that the complex, three-dimensional nature of upper extremity motion analysis is now becoming more feasible’,*. Knowledge of the normal and pathological kinematics is essential for evaluation of motor control3 and perceptual-motor abilities4 and development of elbow joint prostheses and their postoperative therapy5-7. Because the primary function of the upper extremity is to place the hand in space, true evaluation must occur during activities with real life requirements8. A ‘biomechanical profile’ of specific tasks is needed in order to improve performance and minimize risk of injuryg. Sophistication
and
Received: 9 May 1992 Accepfed: 17 December 1992 Correspondence and reprint requests to: Tanya L Packer, Division of
Occupational Therapy, School of Rehabilitation Therapy, Louise D Acton Bldg, Queen’s University, Kingston, Ontario, Canada, K7L 3N6 @ 1993 Butterworth-Heinemann 0268~0033/93/060315-07
Ltd
Electrogoniometry has been used to report the elbow3,10 and wrist6,‘l motions required during functional activities. However, little or no reliability or validity studies appear to substantiate any of the findings. While reports of photographic techniques to measure gait abound in the literature, very few reports of upper extremity applications exist. Engen and Spencer’* used 35mm photography to record upper limb movement during functional activities, producing descriptive results. Validation of the system was not reported. One group of researchers has used a video-based system to evaluate range of motion during feeding tasks under normal conditions’ and with elbow movement restricted13. Mean maximum error in calculating the coordinates from static marker data did not exceed 1.53% (SD, 0.34) in any coordinate. Dynamic error (n = 10) ranged from -0.85 to 2.65% 14. The error in degrees was not reported. Optoelectric systems use active markers that eliminate difficulties with trajectory crossings and the need for manual digitization15. Spatial resolution is better than 60 Hz standard video systems’. There are
316
C/in. Biomech.
1993; 8: No 6
also difficulties: the cabling system necessitates testing in controlled environments and reflections from surrounding objects cause distortions’. Two studies1,16 have examined the reliability and validity of the WATSMART@ (Waterloo Spatial Motion Analysis Recording Technique) system. In the first study, calculated angles compared to a preset goniometer and a robot arm at various spatial heights and rotations were within 0.5”. Reliability decreased slightly as the angles increased and as the orientation to the camera was rotated 45”. There was no evidence of a proportional error, but there was a constant error with all angles calculated as slightly less (0.4-1.4”) than the reference angle. The author concluded that movement trajectories can be reconstructed reliably using this optoelectric system’. These results were specific to the set-up described by the authors and were influenced by such factors as camera angles, camera distance from the movement being measured, light reflections and marker visualization. While they are encouraging, any generalization must be made with caution. Deluzio et al. l6 also investigated the accuracy of the WATSMAR? system. They used both a static and a dynamic calibration dummy that mimicked the lower extremity. In all cases the flexion error was less than 1” and rotation error was less than 2”. Adduction error was slightly larger; however, this measurement represents the carrying angle in upper extremity studies and is of little significance in functional movements. Adding the complexity of 3 degrees of freedom during the dynamic trials did not affect accuracy16. This paper will report on application of a set-up, similar to that used by Deluzio et al.16, to measurement of the elbow with the assumption that measurement during functional activity is the projected goal. The purposes of the paper are: (1) to review types and accuracy of measurement systems previously used to measure elbow motion during functional activities; (2)
to confirm accuracy of the upper extremity application with the calibration dummy; and (3) to report the test-retest reliability with human subjects. Methods
Differences between this set-up and the one used by Deluzio et a1.r6 were the room, LED configuration, camera placement, and reference position protocol which did not include the use of X-rays. Because of the extensive testing previously performed, only static and one dynamic test, using the calibration dummy configured as the upper extremity, were carried out. Following this confirmation of accuracy, test-retest studies were carried out with human subjects. Ethical approval for all testing on human subjects was granted through the School of Rehabilitation Therapy, Queen’s University, and all subjects signed an informed consent form prior to testing. Calibration Calibration of optoelectric system
The two cameras were placed in a uniform location with respect to all three planes, pointing toward the testing area; the angle between them was 55” and the cameras were approximately 2 m away from the testing area. The known problems with reflections were minimized with the use of black carpet, drapes, and ceiling tiles. Once positioned, the testing area was calibrated using the frame supplied with the system. Calibration the parameters calculated needed for the reconstruction equations (direct linear transformation algorithm) used to change the combined 2-D data from the two cameras to 3-D coordinates. Calibration of the reference LEDS
Figure 1. Laboratory set-up with subject wearing outriggers and placed in known position relative to reference positioning table. Note: the black curtain hung behind the testing area was omitted for photographic clarity.
Placed to form three non-collinear points in each segment (arm and forearm), the LEDs were attached to outriggers moulded from thermoplastic (Orthoplast, Johnson and Johnson Co.). One common LED, placed over the lateral epicondyle of the elbow, was used (Figure 1). These defined each segment as a rigid body in 3-D space. The outriggers were attached to the skin using a combination of electrode collars and 2” wide elastic. The calculation of joint angles from the marker data required the ability to identify the location of the defined axis of rotation (Figure 2). Two data files with known reference positions were collected. The first defined a plane on the reference positioning table. In the second reference position the subject was positioned so that the defined anatomical axes of rotation for elbow flexioniextension and forearm rotation were in known positions relative to the first reference position and relative to clinically defined angles (90” flexion, 0” rotation and 0” ad/abduction). This positioning was critical; any inaccuracy was
Packer et al.: Optoelectric system and elbow kinematics randomly in three planes, measuring the angles.
again
317
simultaneously
Test-retest reliability
zft Figure 2. Location of floating axis coordinates. Fixed body axes = Y, and Zr; Reference axes = Z, and Yf; Orthogonal axes =X, and Xr; X, = line through centre of rotation in a posteroanterior direction; Y, = line through trochlea perpendicular to humerus; Z, = line through shaft of humerus; X,= line through the distal heads of the radius and ulna; Y, = line through wrist perpendicular Xr; Z,= line through forearm from elbow to wrist.
magnified by the calculations used to determine relative angles. Positioning was done with an inclinometer, manual and gravity reference goniometers, and positioning lines on the reference table. Although the location of the defined axes of rotation were known in the second reference position, data was collected from the LEDs on the outriggers attached to the subject. Because the locations of the anatomically defined axes of rotation and the LEDs relative to the reference table markers were both known, the difference between the LED location and the anatomical axes could be calculated. This known relationship was used for all successive calculations so that localization of the defined axes in space was always possible. The floating axis method17 was used to calculate clinical angles. In order to calculate relative angles three coordinate systems were defined: one in each body segment (fixed body axes) and one at the joint centre of rotation. The joint coordinate system used the two fixed body axes and a floating axis; a common perpendicular to the other two. It was not fixed to either body and moved in relation to both (Figure 2). Experimentation Calibration dummy The calibration dummyn’ was statically set at 10 intervals of flexion from 0 to 110” (total = 12 positions) and at four rotation angles with the elbow at approximately 0, 45, and 90” of flexion. Angles were recorded simultaneously with the potentiometers of the calibration dummy and the optoelectric system. During the dynamic trial the calibration dummy was moved
The test-retest protocol consisted of instrumentation of eight female subjects twice each, including calibration of the system and collection of both reference positions. Mean age was 34.5 years (SD = 8.3). For each subject the order of collection was randomized for test 1 and test 2 individually. Time between trials was approximately 2 min and between test 1 and test 2 approximately 10 min. All instrumentation and positioning was carried out by the same investigator. Subjects were positioned in nine flexion/extension angles and five pronation/supination angles using a gravity reference goniometer, an inclinometer, and a two-arm manual goniometer to standardize rotation, arm abduction, and elbow flexion./extension. Static joint measurement in clinical settings is most frequently performed using the manual goniometer. Using intra-class correlations (ICC) intra-tester reliability reports have ranged from 0.91 to 0.99 ‘sY2”. Inter-tester reliability is lower and more varied than intra-tester reliability18~20~2’with reports as low as 0.26 for some shoulder movements2’. Reports at the elbow have ranged from 0.95 to 0.97 for flexion/extension’“. Variation in measurements in clinical settings reported in degrees have yielded standard deviations of 4”22 and 4.8021. In a controlled laboratory setting elbow measurement accuracy was +2.4-3.4”*“, slightly better than in the clinical setting. At each angle the difference between test 1 and test 2 measurements and the least significant difference (LSD) were calculated. The LSD is an estimate of the difference between two readings from matched pairs (assuming the difference is normally distributed) and is calculated using the formula LSD = 1 x SD (where t = the value for t at the desired significance level). A two-tailed, 95% confidence level was chosen. With measured differences below the calculated LSD there is insufficient evidence to conclude that the two values are different24. Results Calibration dummy Results of tests conducted with the calibration dummy are shown in Tables 1 to 3. When the measured potentiometer angles and those computed by the optoelectric system during flexion’extension were compared (Table l), the overall mean difference was less than -0.25” (SD = 0.39”). It can be seen that as the flexion angle decreased (approached extension) the agreement between the two measurement systems decreased slightly. Table 2 illustrates the difference between the rotation measured with the potentiometers and that calculated by the optoelectric system. As the forearm
318
Clin. Biomech.
1993; 8: No 6
Table 1. Comparison of measured and computed flexion/ extension angles (deg) Measured
Computed
Difference
0.8 11.6 20.9 31.8 42.1 50.8 61.3 71.8 81.4 90.6 101.5 111.9
-0.9 -0.8 -0.4 -0.3 -0.0 0.4 0.2 -0.2 -0.4 -0.5 -0.1 0.1
-0.1 10.8 20.5 31.5 42.1 51.2 61.5 71.6 81 .O 90.1 101.4 112.0
Mean SD
-0.24 0.39
Measured, potentiometer angle (deg) on calibration dummy. Computed, optoelectric system computed angles (deg). Difference, measured-computed. SD, standard deviation.
was rotated away from neutral, the difference increased. Four different rotation angles, measured at three flexion angles yielded an overall mean error of 1.6” (SD = 1.5”). Abduction/adduction angles (carrying angle) were also computed by the optoelectric system at different flexion angles. The difference between the measured and computed abduction/adduction values varied from 2.4 to 0.0” as the elbow moved from approximately 0” (complete extension) to 110” of flexion. The overall mean error was 0.9” (SD = 1.0’) (Table 3). A dependence on the degree of flexion was seen. Figure 3 demonstrates the similarity between the potentiometer and the LED calculated data during a dynamic trial.
Table 3. Fluctuation in abduction/adduction angle) angles with different flexion angles AbductionIadduction
(carrying
angles (deg)
Flexion angle
Measured
Computed
Difference
-0.1 10.8 20.5 31.5 42.1 51.2 61.5 71.6 81 .O 90.1 101.4 112.0
-0.5 -0.6 -0.6 -0.6 -0.6 -0.6 -0.6 -0.6 -0.6 -0.6 -0.6 -0.6
-2.9 -2.9 -2.7 -2.3 -1.9 -1.6 -1.2 -0.5 -0.8 -0.6 -0.6 -0.3
2.4 2.0 2.1 1.7 1.3 1.0 0.6 -0.1 0.3 0.0 0.0 -0.3 Mean SD
0.9 1.0
Measured, potentiometer angle (deg) on calibration dummy. Computed, optoelectric system angles (deg). Difference, measured-computed. SD, standard deviation.
ior 5 0 -5 -10 -15 -20 -25
a
w I
I
I
I
I
I
I
I
I
I
80r
Table 2. Comparison of measured and computed pronationkupination angles at different flexion angles Pronation/supination
angles (deg)
Flexion angle
Measured
Computed
Difference
-0.6 -0.7 -0.6 -0.6 45.8 45.7 45.6 45.7 90.2 90.3 90.2 90.2
-13.3 2.1 12.1 30.4 -14.3 1.0 16.8 29.7 -14.3 1.7 15.4 30.1
-15.8
2.5 1.0 2.8 3.0 1.2 -0.9 1.6 2.8 -0.5 0.0 1.5 3.7
;*; 2714 -13.1 - 1.9 15.2 26.9 -13.8 1.7 13.9 26.4 Mean SD
1.6 1.5
Measured, potentiometer angle (deg) on calibration dummy. Computed, optoelectric system angles (deg). Difference, measured-computed. SD, standard deviation.
-20 1
I
I
-20 1 0
I 2
I
I
I
I
I
4
6
8
I
I
10
Time (s)
Figure 3. Comparison of measured and computed angles during a dynamic trial. a, Adduction (positive); b, flexion (positive); c, supination (positive). Measured (-----) = potentiometer angles (deg) on calibration dummy. Computer (-_) = optoelectric system computed angles
Meg).
Packer
Test 1
Test 2
so
Mean
Mean
SD
@cd
(ded
Flexion angles 15 30 45 60 75 90 105 120 135
20.7 31.2 46.2 60.1 75.0 90.0 105.4 120.1 136.1
4.1 3.8 4.2 4.2 3.1 3.2 2.9 3.6 4.0
22.6 32.9 46.9 62.1 75.6 90.4 107.0 121.5 135.3
4.6 2.8 3.8 2.8 2.2 2.5 1.7 3.9 2.6
Rotation angles -45 -20 0 20 45
-40.9 -17.9 -1.1 16.3 45.4
7.2 6.4 3.5 5.0 5.9
-38.8 -17.5 -0.6 18.5 44.4
8.0 4.4 5.2 3.9 3.7
Discussion
Upper extremity motion analysis presents several challenges. These include continual and correct visualization of the markers, establishing the reference position relative to both clinical convention and anatomical axes of rotation, and lack of a ‘gold standard’ to which comparisons can be made. The procedure for establishing reference positions in the reported system had the advantage that the LEDs did not have to be placed over bony landmarks, nor did X-rays need to be taken. However, placement of the arm in the reference position was critical. As in any link segment system, movement of the proximal links resulted in movement in the distal segments. Placement of the arm in adduction was crucial; shoulder abduction of even 2” resulted in a change in forearm rotation. In order for the calculated angles to correspond to clinical angles, this position had to be r’eplicated on the reference table. Testing with the calibration dummy has shown the mean error to be less than 1.6” in flexion/extension, rotation, and abduction/adduction for the upper extremity configuration. This is slightly better than the
* Positive angle, supination; negative angle, pronation.
Test-retest
reliability
Table 4 displays the results of the flexion/extension and rotation angles as well as the goniometer settings for test 1 and test 2. The term ‘difference’ was used to describe the difference between the manual goniometer setting and the WATSMART@ data. It should be remembered that this difference includes discrepancy due to inaccuracy in both measurement techniques. The mean difference between the two tests averaged from -2.3” to 1” (Table 5). The LSD values (4.9-13.2 for llexion and 10.7-19.4” for rotation) indicated less Table 5. Difference
and least significant difference
between test 1 and test 2 Subject
1
2
Flexion angles 15.0 30.0 45.0 60.0 75.0 90.0 105.0 120.0 135.0
0.3 2.1 2.0 0.5 -1.1 -0.5 -0.3 -3.4 1.5
-2.2 -4.9 0.8 -5.3 0.5 -1.8 -3.2 -5.1 1.1
Rotation angles -45.0 -20.0 0.0 20.0 45.0
-1.3 -0.1 -1.6 0.0 0.6
2.0 -3.6 3.6 -8.1 -3.4
t0.025for 8 - 1 = 2.3646. LSD, least significant difference. SD, standard deviation.
3
319
rotation tests than the the agreement in flexion/extension tests. Rotation angles also showed increased LSD values as the forearm was rotated further away from neutral. This was more evident as it moved into pronation (LSD = 19.4”) than supination (LSD = 15’). At 60” of flexion there was a larger mean and standard deviation than at any other flexion angle. The LSD indicated that differences of greater than 13” would indicate a significant difference between these two measurements using this WATSMAR'I@ set-up. At other angles the difference required is as small as 4.9”.
Table 4. Results of test 1 and test 2: mean and standard deviation of calculated angles throughout the flexion and rotation ranges Flexion and rotation angles* (degl
et al.: Optoelectric system and elbow kinematics
4
5
6
1.8 1.0 2.4 4.4 1.0 5.2 3.2 7.5 3.6
-0.4 2.2 4.9 5.3 3.3 0.2 -1.3 0.5 2.9
0.9 -4.3 -4.2 -2.0 -0.6 -0.4 -0.7 -1.6 2.7
-7.7 -5.6 -1.7 -7.4 -0.4 -1.1
5.2 10.5 7.3 -2.3 14.9
8.4 -0.6 -1.0 -1.1 4.8
-0.2 -1.8 -0.5 -2.7 -1 .o
-14.3 -2.9 -3.5 3.6 -5.1
7
8
Mean
SD
LSD
-1 .o -5.1 -2.6 -10.5 -0.7 -6.5 1.2 -1.2 -2.7
-4.4 -4.9 -5.6 -0.8 -1.4 2.1 -4.6 -7.2 -1.6
-0.7 -1.7 -0.7 -2.0 -0.6 -0.4 -1.6 -1.4 0.8
2.1 3.4 3.7 5.6 2.5 3.4 3.4 4.4 2.3
4.9 7.9 8.8 13.2 6.0 7.9 7.9 10.4 5.5
-13.9 -4.4 -8.0 -13.5 -0.9
-2.0 0.1 -0.4 6.1 -1.8
-2.0 -0.4 -0.5 -2.3 1.0
8.2 4.7 4.5 6.2 6.3
19.4 11.1 10.7 14.7 15.0
320
Clin. Biomech.
1993; 8: No 6
reported error for the lower extremity configuration16. Although, abduction/adduction measurement seemed to show a dependence on flexion angle, it was not considered critical, since carrying angle is not a significant factor during functional movement. The combined evidence has led us to conclude that the system error prior to application to a human subject is within acceptable limits. In the test-retest experiments the larger LSD values found during rotation indicate that larger differences between groups would be required to make conclusions on differences in technique or pathological groups using this set-up. This is particularly true for values in pronation. This decrease in accuracy with rotation away from the camera was also reported by Scholz’. When using the system ‘in vivo’ additional error could arise from several sources; anatomical/structural differences between the calibration dummy and the human subjects; anatomical differences between individual subjects; poor positioning of the subject in the reference position during calibration, and skin and soft-tissue movement under the attachment of the LED outrigger. These may individually or collectively lead to poorer results. While the ‘gold standard’ during the mechanical testing (potentiometer) has been shown16 to be within l”, the only available ‘gold standard’ for in-vivo measurement (manual goniometry) is known” to be reliable only to k4.8”. In addition there are no reported studies of the validity of clinical goniometric measurement compared to anatomical changes in angle. There is an unknown relationship between clinical convention and anatomy. Use of the manual goniometer to set the arm position in both test 1 and test 2 must account for some of the test-retest difference. Many of the LSD values recorded were less than the variability of the goniometer. In addition the protocol required that subjects hold each angle for 5-10 s after being set with the goniometers. Neither subject nor investigator would be able to detect small changes in angle during that time period. The relatively higher LSD values at 60” of flexion remains unexplained. This was not found to be less reliable using the calibration dummy. Perhaps subjects found this an awkward or uncomfortable position to hold. Although it is impossible to calculate the contribution of the goniometer error and possible subject movement, the high reliability of the system with the calibration dummy and the low test-retest reliability of manual goniometry, suggest that these findings represent a very conservative estimate of the ‘in-vivo’ reliability. Conclusion
The described system has test-retest reliability within acceptable limits for investigation of functional movement of the elbow. Repeatability ranged from 4.9 to 13.6” for flexion angles and 10.7 to 19.4” for rotation
angles. Functional motion of the elbow can be undertaken within this understanding of the system.
Acknowledgements
The authors would like to thank Gerry Saunders for technical advice and assistance throughout this project. We would also like to thank the Physicians’ Services Incorporated who provided partial funding for the project. Finally we would like to thank those individuals who donated their time to be subjects without whom this study could not have been completed.
References 1 Scholz J. Reliability and validity of the WATSMART three-dimensional optoelectric motion analysis system. Phys 7%. 1989; 69 (8): 679-89 2 Daleiiden SK, Fetters L. Kinematic analysis of reaching by active elders: A preliminary report. Phys Ther 1988; 68: 810 3 Segal RL, Youssef EL. Wrist movements in able-bodied and brain injured individuals. Arch Phys Med Rehabil 1991; 72: 454-9 4 Stillman B. Computer-based video analysis of movement. Aust Physiother 1991; 37 (4): 219-27 5 Brumfield RH, Champoux JA. A biomechanical study of normal functional wrist motion. Clin Orthop Rel Res 1984; 187: 23-5 6 Palmer AK, Werner FW, Murphy D, Glisson R. Functional wrist motion: A biomechanical study. J Hand Surg 1985; lOA(1): 39-46
7 Packer TL, Peat M, Wyss UP, Sorbie C. Examining the elbow during functional activities. Occup Ther J Res 1990; lO(6): 323-33 8 Safaee-Rad R, Shwedyk E, Quanbury AO, Cooper JE. Normal functional range of motion of upper limb joints during performance of three feeding activities. Arch Phys Med Rehabill990; 71: 505-9 9 An K-N, Bejjani FJ. Analysis of upper-extremity performance in athletes and musicians. Hand Clinics 1990; 6(3): 393-403
10 Morrey BF, Askew LJ, An K-N, Chao EY. A biomechanical study of normal functional elbow motion. J Bone Joint Surg 1981; 63A (6): 872-7
11 Cooke JD, Brown SH, Cunningham DA. Kinematics of arm movements in elderly humans. Neurobiol Aging 1989; 10: 159-65 12 Engen TJ, Spencer WA. Method of kinematic study of normal upper extremity movements. Arch Phys Med Rehabill968; Jan: 9- 12 13 Cooper, B, Quanbury A. Three-dimensional analysis of the effect of elbow joint restriction on upper limb motion during performance of three feeding activities. Proceedings of the International Symposium on 3-D Analysis of Human Movement 1991; July: (Abstract)
14 Safaee-Rad R, Schwedyk E, Quanbury AO. Three-dimensional measurement system for functional arm motion study. Med Biol Eng Comput 1990; 28: 569-73 15 Rowe11 D, Mann RW. Human movement analysis. SOMA Eng Hum Body 1989; 3(2): 13-20
Packer et al.: Optoelectric system and elbow kinematics
16 Deluzio KJ, Wyss UP, Li J, Costigan PA. A procedure to
validate three-dimensional motion assessment systems. .I Biomech (in press) 17 Grood ES, Suntay WJ. A joint coordinate system for the clinical description of three-dimensional motions: Application to the knee. J Biomech Erg 1983; 105: 136-44 18 Horger MM. The reliability of goniometric measurements of active and passive wrist motions. Am J Occup Ther 1990; 44(4): 342-8 19 Rothstein JM, Miller PJ, Roettger RF. Goniometric reliability in a clinical setting. Phys Ther 1983; 63 (10): 1611-15
321
20 Riddle DL, Rothstein JM, Lamb RL. Goniometric reliability in a clinical setting. Phys Ther 1987; 67(5): 668-73 21 Bovens AMP, van Baak MA, Vrencken JGPM et al. Variability and reliability of joint measurements. Am J Sports Med 1990; 18 (1): 58-63 22 Boone DC, Azen SP, Lin CM et al. Reliability of goniometric measurements. Phys Ther 1978; 58: 1355-60 23 Fish DR, Wingate L. Sources of goniometric error at the elbow. Phys Ther 1985; 65(11): 1666-70 24 Tillotson KM, Burton, AK. Noninvasive measurement of lumbar sagittal mobility: An assessmen: of the flexicurve technique. Spine 1991; 16(l): 29-33
1993; 8: 315-321
Reliability of an optoelectric measure elbow kinematics
system to
T L Packer
MSc’,
PhD’,
U P Wyss Pm*, P A Costigan
‘Department of Anatomy and Division of Occupational Group, Queen’s University, Kingston, Ontario, Canada
Therapy, and ‘Clinical
Mechanics
Summary Motion analysis of the upper extremity during functional activities has only recently become more accessible. An optoelectric system (WAXMART) was subjected to testing using first a calibration dummy and then human subjects. The mean differences between the system calculated angles and those measured with potentiometers on a calibration dummy were less than 1.6” in all three planes. The test-retest reliability of the system when measuring elbow motion of human subjects yielded acceptable repeatability for measurement of functional activities. Calculation of the least significant difference found that minimum differences from 4 to 14” in flexion and from 10 to 19” in rotation can be detected using the reported set-up and protocol.
Relevance Understanding of the kinematic demands of activities of daily living will aid in the development of elbow joint prostheses and subsequent postoperative therapy. Evaluation of motor control and perceptual-motor abilities and treatment of pathological conditions will thus ensue. The first step in this understanding is knowledge of the limits of the measurement tool. This paper sets out these limits. Key words: Upper extremity, motion analysis, optoelectric, activities of daily living, elbow joint, three-dimensional
Introduction
development of technology and techniques has improved in recent years so that the complex, three-dimensional nature of upper extremity motion analysis is now becoming more feasible’,*. Knowledge of the normal and pathological kinematics is essential for evaluation of motor control3 and perceptual-motor abilities4 and development of elbow joint prostheses and their postoperative therapy5-7. Because the primary function of the upper extremity is to place the hand in space, true evaluation must occur during activities with real life requirements8. A ‘biomechanical profile’ of specific tasks is needed in order to improve performance and minimize risk of injuryg. Sophistication
and
Received: 9 May 1992 Accepfed: 17 December 1992 Correspondence and reprint requests to: Tanya L Packer, Division of
Occupational Therapy, School of Rehabilitation Therapy, Louise D Acton Bldg, Queen’s University, Kingston, Ontario, Canada, K7L 3N6 @ 1993 Butterworth-Heinemann 0268~0033/93/060315-07
Ltd
Electrogoniometry has been used to report the elbow3,10 and wrist6,‘l motions required during functional activities. However, little or no reliability or validity studies appear to substantiate any of the findings. While reports of photographic techniques to measure gait abound in the literature, very few reports of upper extremity applications exist. Engen and Spencer’* used 35mm photography to record upper limb movement during functional activities, producing descriptive results. Validation of the system was not reported. One group of researchers has used a video-based system to evaluate range of motion during feeding tasks under normal conditions’ and with elbow movement restricted13. Mean maximum error in calculating the coordinates from static marker data did not exceed 1.53% (SD, 0.34) in any coordinate. Dynamic error (n = 10) ranged from -0.85 to 2.65% 14. The error in degrees was not reported. Optoelectric systems use active markers that eliminate difficulties with trajectory crossings and the need for manual digitization15. Spatial resolution is better than 60 Hz standard video systems’. There are
316
C/in. Biomech.
1993; 8: No 6
also difficulties: the cabling system necessitates testing in controlled environments and reflections from surrounding objects cause distortions’. Two studies1,16 have examined the reliability and validity of the WATSMART@ (Waterloo Spatial Motion Analysis Recording Technique) system. In the first study, calculated angles compared to a preset goniometer and a robot arm at various spatial heights and rotations were within 0.5”. Reliability decreased slightly as the angles increased and as the orientation to the camera was rotated 45”. There was no evidence of a proportional error, but there was a constant error with all angles calculated as slightly less (0.4-1.4”) than the reference angle. The author concluded that movement trajectories can be reconstructed reliably using this optoelectric system’. These results were specific to the set-up described by the authors and were influenced by such factors as camera angles, camera distance from the movement being measured, light reflections and marker visualization. While they are encouraging, any generalization must be made with caution. Deluzio et al. l6 also investigated the accuracy of the WATSMAR? system. They used both a static and a dynamic calibration dummy that mimicked the lower extremity. In all cases the flexion error was less than 1” and rotation error was less than 2”. Adduction error was slightly larger; however, this measurement represents the carrying angle in upper extremity studies and is of little significance in functional movements. Adding the complexity of 3 degrees of freedom during the dynamic trials did not affect accuracy16. This paper will report on application of a set-up, similar to that used by Deluzio et al.16, to measurement of the elbow with the assumption that measurement during functional activity is the projected goal. The purposes of the paper are: (1) to review types and accuracy of measurement systems previously used to measure elbow motion during functional activities; (2)
to confirm accuracy of the upper extremity application with the calibration dummy; and (3) to report the test-retest reliability with human subjects. Methods
Differences between this set-up and the one used by Deluzio et a1.r6 were the room, LED configuration, camera placement, and reference position protocol which did not include the use of X-rays. Because of the extensive testing previously performed, only static and one dynamic test, using the calibration dummy configured as the upper extremity, were carried out. Following this confirmation of accuracy, test-retest studies were carried out with human subjects. Ethical approval for all testing on human subjects was granted through the School of Rehabilitation Therapy, Queen’s University, and all subjects signed an informed consent form prior to testing. Calibration Calibration of optoelectric system
The two cameras were placed in a uniform location with respect to all three planes, pointing toward the testing area; the angle between them was 55” and the cameras were approximately 2 m away from the testing area. The known problems with reflections were minimized with the use of black carpet, drapes, and ceiling tiles. Once positioned, the testing area was calibrated using the frame supplied with the system. Calibration the parameters calculated needed for the reconstruction equations (direct linear transformation algorithm) used to change the combined 2-D data from the two cameras to 3-D coordinates. Calibration of the reference LEDS
Figure 1. Laboratory set-up with subject wearing outriggers and placed in known position relative to reference positioning table. Note: the black curtain hung behind the testing area was omitted for photographic clarity.
Placed to form three non-collinear points in each segment (arm and forearm), the LEDs were attached to outriggers moulded from thermoplastic (Orthoplast, Johnson and Johnson Co.). One common LED, placed over the lateral epicondyle of the elbow, was used (Figure 1). These defined each segment as a rigid body in 3-D space. The outriggers were attached to the skin using a combination of electrode collars and 2” wide elastic. The calculation of joint angles from the marker data required the ability to identify the location of the defined axis of rotation (Figure 2). Two data files with known reference positions were collected. The first defined a plane on the reference positioning table. In the second reference position the subject was positioned so that the defined anatomical axes of rotation for elbow flexioniextension and forearm rotation were in known positions relative to the first reference position and relative to clinically defined angles (90” flexion, 0” rotation and 0” ad/abduction). This positioning was critical; any inaccuracy was
Packer et al.: Optoelectric system and elbow kinematics randomly in three planes, measuring the angles.
again
317
simultaneously
Test-retest reliability
zft Figure 2. Location of floating axis coordinates. Fixed body axes = Y, and Zr; Reference axes = Z, and Yf; Orthogonal axes =X, and Xr; X, = line through centre of rotation in a posteroanterior direction; Y, = line through trochlea perpendicular to humerus; Z, = line through shaft of humerus; X,= line through the distal heads of the radius and ulna; Y, = line through wrist perpendicular Xr; Z,= line through forearm from elbow to wrist.
magnified by the calculations used to determine relative angles. Positioning was done with an inclinometer, manual and gravity reference goniometers, and positioning lines on the reference table. Although the location of the defined axes of rotation were known in the second reference position, data was collected from the LEDs on the outriggers attached to the subject. Because the locations of the anatomically defined axes of rotation and the LEDs relative to the reference table markers were both known, the difference between the LED location and the anatomical axes could be calculated. This known relationship was used for all successive calculations so that localization of the defined axes in space was always possible. The floating axis method17 was used to calculate clinical angles. In order to calculate relative angles three coordinate systems were defined: one in each body segment (fixed body axes) and one at the joint centre of rotation. The joint coordinate system used the two fixed body axes and a floating axis; a common perpendicular to the other two. It was not fixed to either body and moved in relation to both (Figure 2). Experimentation Calibration dummy The calibration dummyn’ was statically set at 10 intervals of flexion from 0 to 110” (total = 12 positions) and at four rotation angles with the elbow at approximately 0, 45, and 90” of flexion. Angles were recorded simultaneously with the potentiometers of the calibration dummy and the optoelectric system. During the dynamic trial the calibration dummy was moved
The test-retest protocol consisted of instrumentation of eight female subjects twice each, including calibration of the system and collection of both reference positions. Mean age was 34.5 years (SD = 8.3). For each subject the order of collection was randomized for test 1 and test 2 individually. Time between trials was approximately 2 min and between test 1 and test 2 approximately 10 min. All instrumentation and positioning was carried out by the same investigator. Subjects were positioned in nine flexion/extension angles and five pronation/supination angles using a gravity reference goniometer, an inclinometer, and a two-arm manual goniometer to standardize rotation, arm abduction, and elbow flexion./extension. Static joint measurement in clinical settings is most frequently performed using the manual goniometer. Using intra-class correlations (ICC) intra-tester reliability reports have ranged from 0.91 to 0.99 ‘sY2”. Inter-tester reliability is lower and more varied than intra-tester reliability18~20~2’with reports as low as 0.26 for some shoulder movements2’. Reports at the elbow have ranged from 0.95 to 0.97 for flexion/extension’“. Variation in measurements in clinical settings reported in degrees have yielded standard deviations of 4”22 and 4.8021. In a controlled laboratory setting elbow measurement accuracy was +2.4-3.4”*“, slightly better than in the clinical setting. At each angle the difference between test 1 and test 2 measurements and the least significant difference (LSD) were calculated. The LSD is an estimate of the difference between two readings from matched pairs (assuming the difference is normally distributed) and is calculated using the formula LSD = 1 x SD (where t = the value for t at the desired significance level). A two-tailed, 95% confidence level was chosen. With measured differences below the calculated LSD there is insufficient evidence to conclude that the two values are different24. Results Calibration dummy Results of tests conducted with the calibration dummy are shown in Tables 1 to 3. When the measured potentiometer angles and those computed by the optoelectric system during flexion’extension were compared (Table l), the overall mean difference was less than -0.25” (SD = 0.39”). It can be seen that as the flexion angle decreased (approached extension) the agreement between the two measurement systems decreased slightly. Table 2 illustrates the difference between the rotation measured with the potentiometers and that calculated by the optoelectric system. As the forearm
318
Clin. Biomech.
1993; 8: No 6
Table 1. Comparison of measured and computed flexion/ extension angles (deg) Measured
Computed
Difference
0.8 11.6 20.9 31.8 42.1 50.8 61.3 71.8 81.4 90.6 101.5 111.9
-0.9 -0.8 -0.4 -0.3 -0.0 0.4 0.2 -0.2 -0.4 -0.5 -0.1 0.1
-0.1 10.8 20.5 31.5 42.1 51.2 61.5 71.6 81 .O 90.1 101.4 112.0
Mean SD
-0.24 0.39
Measured, potentiometer angle (deg) on calibration dummy. Computed, optoelectric system computed angles (deg). Difference, measured-computed. SD, standard deviation.
was rotated away from neutral, the difference increased. Four different rotation angles, measured at three flexion angles yielded an overall mean error of 1.6” (SD = 1.5”). Abduction/adduction angles (carrying angle) were also computed by the optoelectric system at different flexion angles. The difference between the measured and computed abduction/adduction values varied from 2.4 to 0.0” as the elbow moved from approximately 0” (complete extension) to 110” of flexion. The overall mean error was 0.9” (SD = 1.0’) (Table 3). A dependence on the degree of flexion was seen. Figure 3 demonstrates the similarity between the potentiometer and the LED calculated data during a dynamic trial.
Table 3. Fluctuation in abduction/adduction angle) angles with different flexion angles AbductionIadduction
(carrying
angles (deg)
Flexion angle
Measured
Computed
Difference
-0.1 10.8 20.5 31.5 42.1 51.2 61.5 71.6 81 .O 90.1 101.4 112.0
-0.5 -0.6 -0.6 -0.6 -0.6 -0.6 -0.6 -0.6 -0.6 -0.6 -0.6 -0.6
-2.9 -2.9 -2.7 -2.3 -1.9 -1.6 -1.2 -0.5 -0.8 -0.6 -0.6 -0.3
2.4 2.0 2.1 1.7 1.3 1.0 0.6 -0.1 0.3 0.0 0.0 -0.3 Mean SD
0.9 1.0
Measured, potentiometer angle (deg) on calibration dummy. Computed, optoelectric system angles (deg). Difference, measured-computed. SD, standard deviation.
ior 5 0 -5 -10 -15 -20 -25
a
w I
I
I
I
I
I
I
I
I
I
80r
Table 2. Comparison of measured and computed pronationkupination angles at different flexion angles Pronation/supination
angles (deg)
Flexion angle
Measured
Computed
Difference
-0.6 -0.7 -0.6 -0.6 45.8 45.7 45.6 45.7 90.2 90.3 90.2 90.2
-13.3 2.1 12.1 30.4 -14.3 1.0 16.8 29.7 -14.3 1.7 15.4 30.1
-15.8
2.5 1.0 2.8 3.0 1.2 -0.9 1.6 2.8 -0.5 0.0 1.5 3.7
;*; 2714 -13.1 - 1.9 15.2 26.9 -13.8 1.7 13.9 26.4 Mean SD
1.6 1.5
Measured, potentiometer angle (deg) on calibration dummy. Computed, optoelectric system angles (deg). Difference, measured-computed. SD, standard deviation.
-20 1
I
I
-20 1 0
I 2
I
I
I
I
I
4
6
8
I
I
10
Time (s)
Figure 3. Comparison of measured and computed angles during a dynamic trial. a, Adduction (positive); b, flexion (positive); c, supination (positive). Measured (-----) = potentiometer angles (deg) on calibration dummy. Computer (-_) = optoelectric system computed angles
Meg).
Packer
Test 1
Test 2
so
Mean
Mean
SD
@cd
(ded
Flexion angles 15 30 45 60 75 90 105 120 135
20.7 31.2 46.2 60.1 75.0 90.0 105.4 120.1 136.1
4.1 3.8 4.2 4.2 3.1 3.2 2.9 3.6 4.0
22.6 32.9 46.9 62.1 75.6 90.4 107.0 121.5 135.3
4.6 2.8 3.8 2.8 2.2 2.5 1.7 3.9 2.6
Rotation angles -45 -20 0 20 45
-40.9 -17.9 -1.1 16.3 45.4
7.2 6.4 3.5 5.0 5.9
-38.8 -17.5 -0.6 18.5 44.4
8.0 4.4 5.2 3.9 3.7
Discussion
Upper extremity motion analysis presents several challenges. These include continual and correct visualization of the markers, establishing the reference position relative to both clinical convention and anatomical axes of rotation, and lack of a ‘gold standard’ to which comparisons can be made. The procedure for establishing reference positions in the reported system had the advantage that the LEDs did not have to be placed over bony landmarks, nor did X-rays need to be taken. However, placement of the arm in the reference position was critical. As in any link segment system, movement of the proximal links resulted in movement in the distal segments. Placement of the arm in adduction was crucial; shoulder abduction of even 2” resulted in a change in forearm rotation. In order for the calculated angles to correspond to clinical angles, this position had to be r’eplicated on the reference table. Testing with the calibration dummy has shown the mean error to be less than 1.6” in flexion/extension, rotation, and abduction/adduction for the upper extremity configuration. This is slightly better than the
* Positive angle, supination; negative angle, pronation.
Test-retest
reliability
Table 4 displays the results of the flexion/extension and rotation angles as well as the goniometer settings for test 1 and test 2. The term ‘difference’ was used to describe the difference between the manual goniometer setting and the WATSMART@ data. It should be remembered that this difference includes discrepancy due to inaccuracy in both measurement techniques. The mean difference between the two tests averaged from -2.3” to 1” (Table 5). The LSD values (4.9-13.2 for llexion and 10.7-19.4” for rotation) indicated less Table 5. Difference
and least significant difference
between test 1 and test 2 Subject
1
2
Flexion angles 15.0 30.0 45.0 60.0 75.0 90.0 105.0 120.0 135.0
0.3 2.1 2.0 0.5 -1.1 -0.5 -0.3 -3.4 1.5
-2.2 -4.9 0.8 -5.3 0.5 -1.8 -3.2 -5.1 1.1
Rotation angles -45.0 -20.0 0.0 20.0 45.0
-1.3 -0.1 -1.6 0.0 0.6
2.0 -3.6 3.6 -8.1 -3.4
t0.025for 8 - 1 = 2.3646. LSD, least significant difference. SD, standard deviation.
3
319
rotation tests than the the agreement in flexion/extension tests. Rotation angles also showed increased LSD values as the forearm was rotated further away from neutral. This was more evident as it moved into pronation (LSD = 19.4”) than supination (LSD = 15’). At 60” of flexion there was a larger mean and standard deviation than at any other flexion angle. The LSD indicated that differences of greater than 13” would indicate a significant difference between these two measurements using this WATSMAR'I@ set-up. At other angles the difference required is as small as 4.9”.
Table 4. Results of test 1 and test 2: mean and standard deviation of calculated angles throughout the flexion and rotation ranges Flexion and rotation angles* (degl
et al.: Optoelectric system and elbow kinematics
4
5
6
1.8 1.0 2.4 4.4 1.0 5.2 3.2 7.5 3.6
-0.4 2.2 4.9 5.3 3.3 0.2 -1.3 0.5 2.9
0.9 -4.3 -4.2 -2.0 -0.6 -0.4 -0.7 -1.6 2.7
-7.7 -5.6 -1.7 -7.4 -0.4 -1.1
5.2 10.5 7.3 -2.3 14.9
8.4 -0.6 -1.0 -1.1 4.8
-0.2 -1.8 -0.5 -2.7 -1 .o
-14.3 -2.9 -3.5 3.6 -5.1
7
8
Mean
SD
LSD
-1 .o -5.1 -2.6 -10.5 -0.7 -6.5 1.2 -1.2 -2.7
-4.4 -4.9 -5.6 -0.8 -1.4 2.1 -4.6 -7.2 -1.6
-0.7 -1.7 -0.7 -2.0 -0.6 -0.4 -1.6 -1.4 0.8
2.1 3.4 3.7 5.6 2.5 3.4 3.4 4.4 2.3
4.9 7.9 8.8 13.2 6.0 7.9 7.9 10.4 5.5
-13.9 -4.4 -8.0 -13.5 -0.9
-2.0 0.1 -0.4 6.1 -1.8
-2.0 -0.4 -0.5 -2.3 1.0
8.2 4.7 4.5 6.2 6.3
19.4 11.1 10.7 14.7 15.0
320
Clin. Biomech.
1993; 8: No 6
reported error for the lower extremity configuration16. Although, abduction/adduction measurement seemed to show a dependence on flexion angle, it was not considered critical, since carrying angle is not a significant factor during functional movement. The combined evidence has led us to conclude that the system error prior to application to a human subject is within acceptable limits. In the test-retest experiments the larger LSD values found during rotation indicate that larger differences between groups would be required to make conclusions on differences in technique or pathological groups using this set-up. This is particularly true for values in pronation. This decrease in accuracy with rotation away from the camera was also reported by Scholz’. When using the system ‘in vivo’ additional error could arise from several sources; anatomical/structural differences between the calibration dummy and the human subjects; anatomical differences between individual subjects; poor positioning of the subject in the reference position during calibration, and skin and soft-tissue movement under the attachment of the LED outrigger. These may individually or collectively lead to poorer results. While the ‘gold standard’ during the mechanical testing (potentiometer) has been shown16 to be within l”, the only available ‘gold standard’ for in-vivo measurement (manual goniometry) is known” to be reliable only to k4.8”. In addition there are no reported studies of the validity of clinical goniometric measurement compared to anatomical changes in angle. There is an unknown relationship between clinical convention and anatomy. Use of the manual goniometer to set the arm position in both test 1 and test 2 must account for some of the test-retest difference. Many of the LSD values recorded were less than the variability of the goniometer. In addition the protocol required that subjects hold each angle for 5-10 s after being set with the goniometers. Neither subject nor investigator would be able to detect small changes in angle during that time period. The relatively higher LSD values at 60” of flexion remains unexplained. This was not found to be less reliable using the calibration dummy. Perhaps subjects found this an awkward or uncomfortable position to hold. Although it is impossible to calculate the contribution of the goniometer error and possible subject movement, the high reliability of the system with the calibration dummy and the low test-retest reliability of manual goniometry, suggest that these findings represent a very conservative estimate of the ‘in-vivo’ reliability. Conclusion
The described system has test-retest reliability within acceptable limits for investigation of functional movement of the elbow. Repeatability ranged from 4.9 to 13.6” for flexion angles and 10.7 to 19.4” for rotation
angles. Functional motion of the elbow can be undertaken within this understanding of the system.
Acknowledgements
The authors would like to thank Gerry Saunders for technical advice and assistance throughout this project. We would also like to thank the Physicians’ Services Incorporated who provided partial funding for the project. Finally we would like to thank those individuals who donated their time to be subjects without whom this study could not have been completed.
References 1 Scholz J. Reliability and validity of the WATSMART three-dimensional optoelectric motion analysis system. Phys 7%. 1989; 69 (8): 679-89 2 Daleiiden SK, Fetters L. Kinematic analysis of reaching by active elders: A preliminary report. Phys Ther 1988; 68: 810 3 Segal RL, Youssef EL. Wrist movements in able-bodied and brain injured individuals. Arch Phys Med Rehabil 1991; 72: 454-9 4 Stillman B. Computer-based video analysis of movement. Aust Physiother 1991; 37 (4): 219-27 5 Brumfield RH, Champoux JA. A biomechanical study of normal functional wrist motion. Clin Orthop Rel Res 1984; 187: 23-5 6 Palmer AK, Werner FW, Murphy D, Glisson R. Functional wrist motion: A biomechanical study. J Hand Surg 1985; lOA(1): 39-46
7 Packer TL, Peat M, Wyss UP, Sorbie C. Examining the elbow during functional activities. Occup Ther J Res 1990; lO(6): 323-33 8 Safaee-Rad R, Shwedyk E, Quanbury AO, Cooper JE. Normal functional range of motion of upper limb joints during performance of three feeding activities. Arch Phys Med Rehabill990; 71: 505-9 9 An K-N, Bejjani FJ. Analysis of upper-extremity performance in athletes and musicians. Hand Clinics 1990; 6(3): 393-403
10 Morrey BF, Askew LJ, An K-N, Chao EY. A biomechanical study of normal functional elbow motion. J Bone Joint Surg 1981; 63A (6): 872-7
11 Cooke JD, Brown SH, Cunningham DA. Kinematics of arm movements in elderly humans. Neurobiol Aging 1989; 10: 159-65 12 Engen TJ, Spencer WA. Method of kinematic study of normal upper extremity movements. Arch Phys Med Rehabill968; Jan: 9- 12 13 Cooper, B, Quanbury A. Three-dimensional analysis of the effect of elbow joint restriction on upper limb motion during performance of three feeding activities. Proceedings of the International Symposium on 3-D Analysis of Human Movement 1991; July: (Abstract)
14 Safaee-Rad R, Schwedyk E, Quanbury AO. Three-dimensional measurement system for functional arm motion study. Med Biol Eng Comput 1990; 28: 569-73 15 Rowe11 D, Mann RW. Human movement analysis. SOMA Eng Hum Body 1989; 3(2): 13-20
Packer et al.: Optoelectric system and elbow kinematics
16 Deluzio KJ, Wyss UP, Li J, Costigan PA. A procedure to
validate three-dimensional motion assessment systems. .I Biomech (in press) 17 Grood ES, Suntay WJ. A joint coordinate system for the clinical description of three-dimensional motions: Application to the knee. J Biomech Erg 1983; 105: 136-44 18 Horger MM. The reliability of goniometric measurements of active and passive wrist motions. Am J Occup Ther 1990; 44(4): 342-8 19 Rothstein JM, Miller PJ, Roettger RF. Goniometric reliability in a clinical setting. Phys Ther 1983; 63 (10): 1611-15
321
20 Riddle DL, Rothstein JM, Lamb RL. Goniometric reliability in a clinical setting. Phys Ther 1987; 67(5): 668-73 21 Bovens AMP, van Baak MA, Vrencken JGPM et al. Variability and reliability of joint measurements. Am J Sports Med 1990; 18 (1): 58-63 22 Boone DC, Azen SP, Lin CM et al. Reliability of goniometric measurements. Phys Ther 1978; 58: 1355-60 23 Fish DR, Wingate L. Sources of goniometric error at the elbow. Phys Ther 1985; 65(11): 1666-70 24 Tillotson KM, Burton, AK. Noninvasive measurement of lumbar sagittal mobility: An assessmen: of the flexicurve technique. Spine 1991; 16(l): 29-33