The effect of marker placement deviations on spinal range of motion determined by video motion analysis.

Key Words: motion; Movement; Spine; Tests and measurements, range of motion; Video recording

The assessment of spinal range of motion (ROM) has important implications for people with low back pain, clinicians, and third-party providers. The American Medical Association guidelines for disability use loss of ROM as a criterion for determining percentage of permanent impairment, which is often translated into a dollar amount paid to the patient by his or her insurance company.[1] Several methods of determining spinal ROM are available, including the use of inclinometers, rehabilitation exercise devices, fingertip-to-floor measurements, and sophisticated video motion analysis. The establishment of the reliability and validity of measurements obtained with these systems is critical, given the important decisions being based on the obtained data. There is little data on the reliability of measurements obtained with the more recently developed methods of determining spinal ROM, particularly by use of video motion analysis systems.

Kinematic analysis systems for study of human motion in research and clinical settings are becoming more prevalent.[2-11] These systems allow for quantitative analyses of human motion. Semiautomated kinematic analysis systems allow for operator correction of mislabeled markers or incorrect tracking. Although this process requires time, it is not as time consuming as the older methods of cinematography and hand-digitization.[3-5] Automated systems that require even less time at tracking markers, however, are more restrictive in movements analyzed. One such system, the Motion Analysis SpineTrak[TM](*) system, is an automated motion analysis system that tracks retroreflective markers and calculates ROM and velocity of spinal motion. This system is a two-dimensional motion analysis system that requires the subject to move in set patterns of motion. These patterns are flexion/extension, lateral side bending, and rotation. The decrease in time to track markers is offset by the restrictions of the motion analyzed. If more varied motions are desired, then a semiautomated system must be used.[12]

There are many sources of error in motion analysis systems when analyzing kinematic measurements. Some of these errors can be due to system error, marker placement by one or more investigators, placement of markers on a subject on multiple days, skin movement over bony landmarks during movement, and subject repeatability. Studies[2,3,6,13,14] have shown that various motion analysis systems are accurate in measuring distance and angles as well as showing good within-trial repeatability. Many reports of normative data, especially on gait variables, have been published.[8,13-15] Fewer reports of repeatability have been published.2,13 No studies on the effect of misplacement of markers for repeated trials have been reported. The effect of replacing markers once they are removed is not known. How much error in placement of markers is tolerable? If motion analysis kinematic data are to be used to assess effects of treatment intervention, then marker placement errors must be known.

The purpose of this study was to investigate the effect of marker placement on ROM measurements as calculated using a video motion analysis system. Marker placement differences represent one component of potential measurement error in these systems. In clinical test-retest settings, marker placement error can be obscured by subject variability. To study changes in angles due to marker removal and replacement, a two-dimensional model of the spine was developed. This model was developed to remove the variability associated with human subjects on repeated trials. By constructing a model of the spine that removes these sources of variation, the effect of marker placement changes could be better isolated. We predicted that significant differences in ROM would be observed between the standard marker placement and purposefully "altered" marker placement trials.

Method

Instrumentation

The Motion Analysis SpineTrak[TM] system is a two-dimensional motion tracking and analysis system. For flexion/extension measurements, 8 markers are tracked and angles and velocity are calculated. With lateral side bending, 10 markers are tracked, and with rotation, 4 markers are tracked. This requires a different marker setup for each motion. A single video camera with an 8-mm lens was used to record the images from retroreflective markers at 60 frames per second. The data were recorded at 60 frames per second via a video recorder. The data were digitized at 15 frames per second with a frame-grabber card and analyzed on an IBM-286-compatible computer equipped with a math coprocessor using SpineTrak[TM] software.

Because the system examines motion in two dimensions, a two-dimensional model of the spine was developed (Fig. 1). Development of the model entailed the use of a flexible object that could be expected to bend in the same manner repeatedly. A metal yardstick was attached to a base of support. Wire leads were attached at the top and run through guides on the sides of the base. Washers were attached to the wire leads to stop at set distances when they encountered the guides. Wire leads were used to reduce the chances of stretch, and multiple repetitions were done to ensure the washers and guides were secured in place. By changing marker set and washer position, lateral side bending and flexion/extension could both be simulated (Figs. 2 and 3). Rotation requires a different marker set and model and thus was not tested as part of this study.

Procedure

Lateral side bending. The video camera was placed 133 cm from the base of the spine model and 118 cm from the floor. This placement allowed for maximal size of image in the video monitor without losing markers during motion. These distances are approximately the same as those used with human subjects. Ten 2.5-cm retroreflective disks were attached in the appropriate setup.12 This was labeled the "normal" position. Twenty sets of five repetitions each were recorded. Markers at T-1, T-12, and S-2 were moved up 2.5 cm and down 2.5 cm. Pilot data in our laboratory suggested that measurements obtained by trained physical therapists were reliable within 2.5 cm of marker placement, which served as the basis for our use of 2.5-cm marker deviations. These marker placements resulted in "standard" and "altered" ROM measurements for lumbar and thoracolumbar regions of the spine. The order of spinal levels and direction moved was randomly chosen. At each position, 20 sets of five repetitions each were recorded. The software provided with the device uses five repetitions of each movement in calculating ROM and velocity. Twenty sets of five repetitions each were obtained to represent 20 "subjects." A typical setup of the cameras, subject, and recording equipment is illustrated in Figure 4.

Flexion/extension. The video camera set-up procedure for flexion/ extension was similar to that for the lateral side-bending phase of the study. Eight retroreflective disks were attached to the spine model in the appropriate setup.[12] This initial positioning was again labeled "normal." Twenty sets of five repetitions each were recorded. Marker levels were again randomly moved up and down 2.5 cm. An additional set of markers, located at the greater trochanter, were used to obtain thoracolumbar-pelvic ROM measurements. Twenty sets of five repetitions each were again recorded at each marker setup. Similar to the lateral side-bending phase of the study, this phase resulted in "standard" and "altered" ROM measurements for the lumbar, thoracolumbar, and thoracolumbar-pelvic regions of the spine.

Data Analysis

Each set of five repetitions was processed and analyzed using SpineTrak[TM] software. Because the system is automated, initial marker identification and checking of correct path tracking were all that was needed to be done for data processing. All data sets had correct initial marker identification. Due to tracking errors, several sets required path corrections. This was done, and averages of each data set were printed.

Results

Descriptive statistics for each "standard" and "altered" marker placement were calculated and are presented in the Table. Student's t tests were used to compare each "altered" marker placement measurement with its respective "standard" placement measurement. Although the ROM measurements within each placement (lumbar, thoracolumbar, and thoracolumbar-pelvic) cannot be considered independent because they are calculated using the shared markers, they were analyzed independently to provide reliability data at each measurement. For purposes of clarity, analyses were organized along two major lines: spinal region (lumbar, thoracolumbar, and thoracolumbar-pelvic) and movement (flexion/extension and lateral side bending).

The t tests for the measurements derived from the lumbar region indicated significant differences in ROM for flexion at all altered marker placements (P < .01). For extension, lumbar ROM was significantly different for the S-2 marker in the upward direction and for both T-12 marker alterations (P < .01).

Results of the t tests on differences in flexion/extension ROM from the thoracolumbar region indicated a significant decrease in flexion ROM when the S-2 marker was raised (P < .01). A significant increase in extension ROM was seen when the S-2 marker was lowered (P < .01), and a significant decrease was observed when the T-1 marker was lowered (P < .01).

Thoracolumbar-pelvic flexion ROM was significantly decreased with elevation or lowering of the greater trochanter marker (P < .05, P < .01) and elevation of the T-1 marker (P < .01). For extension, ROM was increased with alteration of the greater trochanter marker in either direction (P < .01).

Lateral side-bending ROM from the lumbar region was significantly different from the "standard" measurement at all "altered" marker placements.

Moving the S-2 marker up or the T-12 marker down resulted in decreased ROM for both left and right lateral side bending (P < .01). Increased ROM was observed by moving the S-2 marker down or the T-12 marker up (P < .01).

Lateral side-bending ROM from the thoracolumbar region was significantly decreased by raising the S-2 marker up (P < .01) for both left and right sides. An increase in left and right side-bending ROM was observed by lowering the S-2 marker (P < .01). Only the right lateral side-bending ROM was affected by moving the T-1 marker down (P < .01).

Discussion

The purpose of this study was to investigate the effect of moving retroreflective markers from their standard anatomical placement on ROM measurements obtained via video motion analysis. To investigate the effects of these marker deviations without the additional variability associated with human subjects, a flexible model of the spine was constructed and used to obtain the ROM measurements. It was felt that knowing the error based solely on marker placement deviations would help to clarify which components of error are significant in the clinical setting.

The results of this study clearly indicate that altering marker placement will affect ROM measurements obtained via video motion analysis systems. The results seem to indicate that lateral side-bending ROM may be more affected than flexion/extension ROM, at least in our model of the spine. The magnitude of these differences was relatively small (ie, 2 [degrees] - 10 [degrees]). Statistical significance was obtained chiefly because the model of the spine used in the study eliminated error variance that would be expected from human subjects. A previous study with humans (PD Robinson and colleagues, unpublished research) indicated that the intrasubject variability exceeded the observed variability in this study. Without removing markers, the standard deviations for test-retest means of ROM measurements obtained for human subjects ranged from 3 to 20 degrees.

Compared with other measurement systems (ie, use of a liquid inclinometer), we believe the level of error from a 2.5-cm marker movement is acceptable. Because of our pilot data and clinical experience, which suggested that experienced therapists are able to reproduce marker placements within the 2.5-cm boundary, we did not analyze larger deviations from our standard placement. Larger deviations from standard landmarks will result in greater ROM deviations.

Conclusion

Although there were statistically significant changes in ROM measurements when markers were moved 2.5 cm, the ROM changes may be too small to have any clinical impact. These results suggest that video motion analysis systems may be used without undue concern for measurement error from the apparatus itself and that relatively small marker "misplacements" will not adversely affect obtained ROM measurements. In collection of subject or clinical data, it may be important to demonstrate that all examiners are able to palpate marker placement positions within 2.5 cm of each other. More data on asymptomatic subjects and variances in repeated trials need to be reported. (*) Motion Analysis Corp, 3617 Westwind Blvd, Santa Rosa, CA 95403.

References

[1] AMA Guide to the Evaluation of Permanent Impairment. Chicago, Ill: American Medical Association; 1989 [2] Vander Linden DW, Carlson SJ, Hubbard RL. Reproducibility and accuracy of angle measurements obtained under static conditions with the Motion Analysis[TM] video system. Phys Ther. 1992;72:300-305. [3] Whittle M. Calibration and performance of a three-dimensional television system for kinematic analysis. J Biomech. 1982;15:185-196. [4] Miller NR, Shapiro R, McLaughlin TM. A technique for obtaining spatial kinematic parameters of segments of biomechanical systems from cinematography data. J Biomech. 1980;13:535-547. [5] Sutherland DH, Hagy JL. Measurement of gait movements from motion picture film. J Bone Joint Surg [Am]. 1972;54:787-797. [6] Stanerson B, Sahrmann SA, Norton BJ. Comparison of gait variables in cerebral palsy pre-and post-selective posterior dorsal rhizotomy. Phys Ther 1991;71(suppl):S69. [7] Wheeler D, Graves J, Miller G, et al. Functional assessment for prediction of lifting capacity. Spine. In press. [8] Winter DA, Quanbury AO, Hobson DA, et al. Kinematics of normal locomotion: a statistical study based on TV data. J Biomech. 1974;7: 479-486. [9] Gracovetsky S, Newman N, Ferron S, Lewis J. Preliminary report: effects of skin shifts on measurements of spinal kinematics made with external markers. In: Proceedings of the Fifth Biannual Conference and Symposium on Canadian Biomechanics, August 1988. 1988:68-69. [10] Alund M, Lursson S. Three-dimensional analysis of neck motion. Spine. 1990;15:87-91. 11 Ebara S, Yamazaki Y, Harada T, et al. Motion analysis of the cervical spine in athetoid cerebral palsy extension-flexion motion. Spine. 1990; 15:1097-1103. [12] Motion Analysis SpineTrak Operator's Manual. Santa Rosa, Calif: Motion Analysis Corp; 1989. [13] Kadaba MP, Ramakrishnan HK, Wooten ME, et al. Repeatability of kinematic kinetic and electromyographic data in normal adult gait. J Orthop Res. 1989;7:849-860. [14] Thurston AJ, Harris JD. Normal kinematics of the lumbar spine and pelvis. Spine. 1983;8: 199-205. [15] Pearcy MJ, Gill JM, Whittle MW, Johnson GR. Dynamic back movement measured using a three-dimensional television system. J Biomech. 1987;20:943-949.