The In Vitro Reliability of the CODA MPX30 as the Basis for a Method of Assessing the In Vivo Motion of the Subtalar Joint

Transcription

1 ORIGINAL ARTICLES The In Vitro Reliability of the CODA MPX30 as the Basis for a Method of Assessing the In Vivo Motion of the Subtalar Joint Ivan Birch, PhD* Kevin Deschamps, MSc Background: The considerable variation in subtalar joint structure and function shown by studies indicates the importance of developing a noninvasive in vivo technique for assessing subtalar joint movement. This article reports the in vitro testing of the CODA MPX30, an active infrared marker motion analysis system. This work represents the first stage in the development of a noninvasive in vivo method for measuring subtalar joint motion during walking. Methods: The in vitro repeatability of the CODA MPX30 system s measurements of marker position, simple and intermarker set angles, was tested. Angular orientations of markers representing the position of the talus and the calcaneus were measured using a purpose-designed marker placement model. Results: Marker location measurements were shown to vary by less than 1.0 mm in all of the planes. The measurement of a 908 anglewasalsofoundtoberepeatableinallofthe planes, although measurements made in the yz plane were shown to be consistently inaccurate (mean, ). Estimation of segmental orientation was found to be repeatable. Estimations of marker set orientations were shown to increase in variability after a coordinate transform was performed (maximum SD, 1.148). Conclusions: The CODA MPX30 was shown to produce repeatable estimations of marker position. Levels of variation in segmental orientation estimates were shown to increase subsequent to coordinate transforms. The combination of the CODA MPX30 and an appropriate marker placement model offers the basis of an in vivo measurement strategy of subtalar joint movement, an important development in the understanding of the function of the joint during gait. (J Am Podiatr Med Assoc 101(5): , 2011) There is increasing acceptance in clinical biomechanics that the foot should not be treated as a rigid segment during movement analysis. 1 This acknowledges the complexity of foot function during walking and the need for a more detailed understanding of the contribution of the individual joints to foot function. 2 Even in the limited number of studies investigating the use of multisegmental models of the foot, the subtalar joint has rarely been investigated in isolation of the other joints, 3-7 the motion of the subtalar joint often being studied in combination with that occurring at the ankle *Faculty of Health and Human Sciences, University of West London, Brentford, England. Division of Musculoskeletal Disorders, University Hospital of Leuven, Leuven, Belgium. Corresponding author: Ivan Birch, PhD, Faculty of Health and Human Sciences, University of West London, Paragon House, Boston Manor Road, Brentford, Middlesex TW8 9GA, England. ( ivan.birch@uwl.ac.uk) joint With the growing body of evidence suggesting that the ankle joint allows multiplanar movement to occur, the ability to discriminate between movement occurring at the ankle and subtalar joints becomes more crucial. The acquisition of detailed knowledge regarding the movement of the subtalar joint is an essential underpinning for all foot-related health-care practice. The in vitro investigations have all shown that considerable variation occurs between individuals in terms of the structure and function of the subtalar joint Such variations underline the importance of being able to assess the movement of the subtalar joint in individuals rather than the development of an overarching description of subtalar joint function. To date, good-quality in vivo data collected during walking are limited to those collected by Arndt et al 3 (n = 3) and Lundgren et al 7 (n = 6), a total of nine study participants. In these 400 September/October 2011 Vol 101 No 5 Journal of the American Podiatric Medical Association

2 two papers, motion data were collected from marker arrays attached to intracortical pins that had been inserted into bones of the leg and foot of live study participants under local anaesthetic infiltration. The data collected, although limited in participant numbers owing to ethical considerations, represent the best in vivo subtalar joint motion data published to date. Although these data are useful, in view of the variation shown by these and previous studies, they cannot be considered to give a definitive picture of the movement of the subtalar joint. Arndt et al 3 noted that future research should include greater participant numbers to present more universally applicable results, whereas the later paper by Lundgren et al 7 reiterated the findings of the earlier papers, reporting that the talocalcaneal joint displayed greater variability than expected in its motion. The use of skin-mounted markers to measure the movement of an underlying bone is problematic owing to the potential discrepancy between the movement of the skin and that of the bone. Their use in the measurement of subtalar joint motion is particularly problematic 23 because the placement of markers in close physical proximity has a tendency to amplify the effect of marker location errors on subsequent angular calculations. These limitations are acknowledged. Nevertheless, in view of the stated need to increase the number of available data sets and the ethical issues associated with an invasive protocol, it is likely to be the only way in which a substantial body of evidence necessary to underpin clinical practice could be collected. Furthermore, in view of the known variability in subtalar joint structure and function, 19-22, 24 the ability to assess the motion of an individual s subtalar joint will always be desirable, particularly in the clinical context. This study reports the in vitro testing of the CODA MPX30 (Charnwood Dynamics Ltd, Leicestershire, England), an active marker motion analysis system capable of collecting data at frequencies of 100 to 800 Hz from up to 56 sensors. The active nature of the markers allows accurate marker identification to be achieved, irrespective of the proximity of adjacent markers. The system is, therefore, particularly suited to motion analysis of the feet, where markers need to be placed close together. The aim of this study was to investigate potential sources of measurement error arising from the CODA MPX30 and a combination of the CODA MPX30 and the particular marker placements that would be used in later work to develop a noninvasive in vivo method of measuring the movements of the subtalar joint during walking. This investigation was necessary to ensure that the source of any errors identified in the later stages of the development of the in vivo method could be correctly identified. Methods The CODA MPX30 used in this investigation consisted of two sensor units, positioned facing each other, 0.35 m above the ground and 7 m apart. The active markers used by the CODA MPX30 consist of an infrared light-emitting diode (Fig. 1) powered and controlled by a separate small driver box that houses the drive circuitry and a rechargeable battery (Fig. 2). The driver boxes receive trigger information from the sensor units (Fig. 3), causing them to flash in rapid succession. This triggering process allows the signal received from each marker to be identified. As a result, the system can identify each individual marker, even when markers are in close physical proximity. The sensor units each contain three photoelectric detector arrays mounted on a rigid frame. In front of each sensor is a mask of light opaque lines. As infrared light enters the sensor unit through three flat windows, the light passes through the mask before reaching each of the photoelectric detector arrays. As a result, shadows are cast on the detector arrays. As the light-emitting diodes move, the locations of the shadows on the detector arrays also move. The shadow patterns are cross-correlated in real time by a digital signal processing card. The masks of the outside two detector arrays produce vertical shadows, and that of the middle sensor produces horizontal shadows. As a result, the two outside sensors are used to assess lateral movement, and Figure 1. CODA MPX30 markers as used in this study. Journal of the American Podiatric Medical Association Vol 101 No 5 September/October

3 Figure 2. CODA MPX30 driver box as used in this study. Figure 3. CODA MPX30 sensor unit. the middle sensor is used to assess vertical movement. The repeatability of the CODA MPX30 system was tested in three ways. First, the repeatability of the system s estimation of marker location was tested. A single marker was placed in the center of the measurement envelope, and data were collected for 5 sec at 200 Hz. Fifty sets of data were collected over 5 days. Unlike camerabased systems, which require a calibration procedure to be performed before each use, the CODA MPX30 has a factory-installed calibration file that is used for all of the data collections. The infrared sensors used by the CODA MPX30 are located in a single scanner unit and, therefore, occupy the same relative position on each occasion. Although more than one scanner can be used, each scanner produces a threedimensional location for each marker. Once the system has been aligned, the scanners operate collaboratively rather than dependently. Second, the accuracy and repeatability of the system s ability to measure simple angular relationships was tested. An aluminum frame with two arms arranged at 908 was used, with two markers attached to each arm. The angle between the two arms was tested by the institution s School of Engineering and was verified as being 908. Data were collected with the frame in four orientations: parallel to the xy plane, parallel to the xz plane, parallel to the yz plane, and oblique to all three planes. Fifty sets of data were collected, each of 1- sec duration at 200 Hz. Third, the repeatability of estimations of angles between sets of markers was tested. Because this investigation represented the first stage of an investigation into the movement of the subtalar joint, the marker placement model used was one designed specifically for this purpose. The talus and the calcaneus were defined as rigid bodies using eight markers. The calcaneus was defined using four markers (Fig. 4A). The x-axis was defined by a line drawn between a posterior medial marker and a posterior lateral marker. To minimize the possible effects of variations in marker positioning, calculation of the orientation of the y-axis was achieved using a virtual marker (a theoretical marker the position of which is determined mathematically from the position of several actual markers), located midway between two anterior calcaneus markers (Fig. 4A). The y-axis was defined as being at right angles to the x-axis in the direction of this virtual midanterior marker. The talus was defined using the lateral and medial talus head markers and a midankle virtual marker (Fig. 4B) located midway between the medial and lateral malleolus markers. This virtual marker was not intended to represent the exact position of the posterior aspect of the talus or to track its movement but to provide a third noncollinear reference marker for the talus, precluding the possibility of undetectable rotations occurring around an axis coinciding with a line between the medial and lateral talus head markers. The x-axis of the talus was defined as a line drawn between the medial and lateral talus markers, and the y-axis was defined as a line drawn at a right angle to the x-axis in the direction of the midmalleolus virtual marker. For the calcaneus and the talus, the mutually orthogonal z-axis was determined by the position of the x- and y-axes. The repeatability with which the CODA MPX30 could calculate the angles between the sets of markers associated with the two bones was tested by placing the markers on a plastic model of a human lower limb (Fig. 5). This allowed the markers to be placed in positions and at separations that replicated as closely as possible those that would occur if the markers had been placed on a human subject but removed the added complexity of any variation in the actual marker positions 402 September/October 2011 Vol 101 No 5 Journal of the American Podiatric Medical Association

4 Figure 4. Definitions of the x- and y-axes of the calcaneus (superior transverse plane view) (A) and the talus (anterior frontal plane view) (B). produced by movements of a human subject. This stage of the study was necessary to ensure that the compound effect of variations in the assessment of individual marker locations was not producing significant effects on subsequent angular data. Fifty sets of data were collected. Each data set was collected over 5 sec at a frequency of 200 Hz. In each of the three trials, data were collected from static rather than moving markers. Unlike camera-based systems, the CODA MPX30 does not track the motion of light entering a lens. Instead, the system collects a series of static estimations of the marker s position, the frequency of which is determined by the operator. The system is, therefore, not subject to the first-order effects of movements experienced by camera-based systems. The most significant second-order effect is the lag between the time at which the position of the first marker and that of the last marker is estimated. This lag of 170 microseconds between marker scans is allowed for in the system software, which backtracks the marker location to the appropriate time interval. was also found to be repeatable in all of the planes, although measurements made in the yz plane were shown to be consistently inaccurate (mean, ) (Table 2). Estimation of segmental orientation was found to be repeatable (Table 3). Estimations of marker set orientations were shown to increase in Results Results of the marker location measurements were shown to vary by less than 1.0 mm in all of the planes (Table 1). The measurement of a 908 angle Figure 5. CODA MPX30 markers placed on a model of the lower limb. Journal of the American Podiatric Medical Association Vol 101 No 5 September/October

5 Table 1. Results from Repeated Measurements of Static Marker Locations z-axis y-axis x-axis Mean (mm) SD (mm) Maximum (mm) Minimum (mm) Range (mm) Coefficient of variation (%) variability after a coordinate transform was performed (maximum SD, 1.148). Discussion These results show that the CODA MPX30 produces repeatable estimations of the absolute position of a marker. The greatest variability was found in measurements along the y-axis. In this study, the y-axis was defined as being the axis running between the two CODA MPX30 sensor units, parallel to the direction of walking to be used in later stages of the study. As indicated by the manufacturers in their own summary of the specifications of the system, the absolute position of a marker along this axis is the most difficult for the system to estimate. Although the x-axis data showed the smallest absolute variation in position (0.19 mm), as a function of the absolute position along the x-axis (mean, 8.04 mm from the intersection of the three axes), it shows the biggest coefficient of variation (35.36%). This was because the small variations in the estimation of the marker s x-axis position represented a proportionately greater percentage of the small x-axis value. In the case of any subsequent study of the movement of the subtalar joint during walking, the accuracy of the estimation of the static absolute position of any particular marker was rarely a critical issue. The Table 2. Results from Repeated Measurements of 908 Angles Oblique In xy Plane In xz Plane In yz Plane Mean (8) SD (8) Maximum (8) Minimum (8) Range (8) Coefficient of variation (%) objective would be to establish the changes in angles between rigid bodies defined by multiple markers. The issue was, therefore, the degree to which the estimation of a marker s position varied even when the marker was static, that is, a roving error. The fact that the data showed a variation in this estimation of less than 1 mm in all three planes is reassuring. However, the effect of a 1-mm variation in the estimation of a marker s position is not constant when attempting to calculate intersegmental angles. A variation in the estimation of a marker s position of 1 mm if the markers are close together has a proportionally much greater effect than if the markers are further apart. The error in the calculation of the 908 angle could be the result of the same effect, the difficulty in accurately estimating the true y-axis location of some markers, ultimately generating an incorrect estimation of the angle produced by the intersection of the lines drawn through them. The effect of such an error would tend to show itself to a greater or lesser degree, depending on the alignment of the frame. When the frame was aligned with the x- and the y-axes, any effect on the angle would be caused by variations in y-axis coordinates of the markers on the part of the frame aligned with the x-axis. When the frame was aligned with the z- and y-axes, the same would be true of the markers on the part of the frame aligned with the z-axis. However, when the frame was aligned with the x- and z-axes, y-axis coordinate variations would affect all of the markers equally. The results showed the greatest range and coefficient of variation of the estimated angle when the frame was aligned with the xy- and yz-axes, as would be expected from this explanation. The overestimation of the 908 angle in the yz Table 3. Results from Repeated Measurements of Intermarker Set Angles Mean SD Coefficient of Variation (%) Talus marker set relative to CODA MPX30 In xy plane (8) In xz plane (8) In yz plane (8) Calcaneus marker set relative to talus marker set Frontal plane (8) Sagittal plane (8) Transverse plane (8) September/October 2011 Vol 101 No 5 Journal of the American Podiatric Medical Association

6 plane was a relatively constant effect and could, therefore, be taken into account in any subsequent studies. In fact, use of the coordinate transform would negate the necessity to do so. Any method based on the relative movement of adjacent segments effectively cancels the error. Variations in the estimations of segmental orientation in relation to the CODA MPX30 s measurement framework were shown to be greatest in the xy plane, although even in this case the SD for the 50 data sets was only Although the absolute variation increased when the process was further complicated by undertaking a coordinate transform, the SD in all of the planes remained below The variability shown was thought to be largely a combination of discrepancies in the y-axis coordinate data caused by the way in which this was established and the amplifying effect on angular data of having markers in close physical proximity. This variability, therefore, represents the total compound error of the combination of the measurement system and the marker placement model. The disproportionate effect of a small absolute error in the estimation of a marker s position on angular data calculated from markers in close physical proximity remains the most significant implication. Conclusions The ability of the CODA MPX30 to produce repeatable estimations of a marker s position was shown to be good, with variations of less than 1.0 mm in all of the planes. Estimations of a known angle were shown to be repeatable in all of the planes, although inaccurate in the yz plane. The level of variation in segmental orientation estimates was shown to increase when a coordinate transform was undertaken. Although the CODA MPX30 and the marker model were shown to produce repeatable results, this does not imply that application of the method in vivo would produce similarly repeatable outcomes. Skin movement remains an important consideration for all movement analysis based on surface markers. Nevertheless, there is general agreement that despite the displacement of skin-mounted markers, the patterns of movement produced by skinmounted and bone-mounted markers show a generally good level of agreement. 6, 23 The limitation of not being able to place three noncollinear markers on the talus is also a consideration. However, use of the midmalleolus virtual marker, located as it is in a position in the talus where the least relative movement is likely to occur between the talus and the leg, offers a suitable way forward. The combination of the CODA MPX30 and an appropriate marker placement model offers the possibility of measuring in vivo subtalar joint movement, an important development in the understanding of the function of the joint during gait, particularly from a clinical perspective. Financial Disclosure: None reported. Conflict of Interest: None reported. References 1. JENKYN TR, NICOL AC: A multi-segmental kinematic model of the foot with a novel definition of forefoot motion for use in clinical gait analysis during walking. J Biomech 40: 3271, PIAZZA SJ: Mechanics of the subtalar joint and its function during walking. Foot Ankle Clin 10: 425, ARNDT A, WESTBLAD P, WINSON I, ET AL: Ankle and subtalar joint kinematics measured with intercortical pins during the stance phase of walking. Foot Ankle Int 25: 357, HAMEL AJ, SHARKEY NA, BUCZEK FL, ET AL: Relative motions of the tibia, talus and calcaneus during the stance phase of gait: a cadaver study. Gait Posture 20: 147, MICHELSON J, HAMEL A, BUCZEK F, ET AL: The effect of ankle injury on subtalar joint motion. Foot Ankle Int 25: 639, NESTER CJ, LIU AM, WARD E, ET AL: In vitro study of foot kinematics using a dynamic walking cadaver model. J Biomech 25: 453, LUNDGREN P, NESTER C, LIU A, ET AL: Invasive in vivo measurement of rear-, mid- and forefoot motion during walking. Gait Posture 28: 93, CORNWALL MW, MCPOIL TG: Three-dimensional movement of the foot during the stance phase of walking. JAPMA 89: 56, RENDALL GC, ABBOUD RJ: The development of a three plane system for the measurement of the ankle/subtalar joint complex in gait. The Foot 9: 31, CARSON MC, HARRINGTON ME, THOMPSON N, ET AL: Kinematic analysis of a multi-segment foot model for research and clinical applications: a repeatability analysis. J Biomech 34: 1299, SMITH R, RATTANAPRASERT U, O DWYER N: Coordination of the ankle joint complex during walking. Hum Mov Sci 20: 447, MACWILLIAMS BA, COWLEY M, NICHOLSON DE: Foot kinematics and kinetics during adolescent gait. Gait Posture 17: 214, CORNWALL MW, MCPOIL TG: Influence of rearfoot postural alignment on rearfoot motion during walking. The Foot 14: 133, LEARDINI A, BENEDETTI MG, BERTI L, ET AL: Rear-foot, midfoot and forefoot motion during the stance phase of gait. Gait Posture 25: 453, Journal of the American Podiatric Medical Association Vol 101 No 5 September/October

7 15. LUNDBERG A, SVENSSON OK, BYLUND C, ET AL: Kinematics of the ankle/foot complex: part 3. Influence of leg rotation. Foot Ankle 9: 304, MCCULLOUGH CJ, BURGE P: Rotational stability of the loadbearing ankle: an experimental study. J Bone Joint Surg Br 62: 460, MICHELSON JD, SCHMIDT GR, MIZEL MS: Kinematics of a total arthroplasty of the ankle: comparison to normal ankle motion. Foot Ankle Int 21: 278, NESTER CJ, FINDLOW AF, BOWKER P, ET AL: Transverse plane motion at the ankle joint. Foot Ankle Int 24: 164, MANTER JT: Movements of the subtalar and transverse tarsal joints. Anat Rec 80: 397, ROOT ML, WEED JH, SGARLATO TE, ET AL: Axis of motion of the subtalar joint. JAPA 56: 149, VAN LANGELAAN EJA: A kinematic analysis of the tarsal joints. Acta Orthop Scand 54 (suppl 204): 1, ISMAN RE, INMAN VT: Anthropometric studies of the human foot. Bull Prosthet Res Spring: 97, WESTBLAD P, HASHIMOTO T, WINSON I, ET AL: Differences in ankle-joint complex motion during the stance phase of walking as measured by superficial and bone anchored markers. Foot Ankle Int 23: 856, NESTER CJ: Lessons from dynamic cadaver and invasive bone pin studies: do we know how the foot really moves during gait? J Foot Ankle Res 2: 18, September/October 2011 Vol 101 No 5 Journal of the American Podiatric Medical Association