Keywords
kinematics - dog - variability
Introduction
Computer-assisted kinematic analysis has been used to characterize gait patterns in
animals with musculoskeletal pathology.[1]–[7] Although there are different kinematic analysis systems available, video-based motion
analysis using superficial skin markers is the most commonly reported method of kinematic
data acquisition of the hindlimb in dogs.[1]–[6],[8],[9]
DeCamp and colleagues described a two-dimensional kinematic model of the hindlimb
in dogs, most commonly used to describe sagittal plane motion. This model has been
widely used to date[1]–[6],[10],[11] and uses superficial skin markers placed over anatomical landmarks for easy identification
and repeatable placement. Nevertheless, previous reports have suggested that inconsistent
marker placement or marker placement error can result invariability in joint kinematics
in the hindlimb.[1],[2],[4],[5] Furthermore, recent research has shown that the direction of marker placement error
can affect the magnitude of detectable differences in kinematic data.[12] What is less apparent are factors leading to erroneous marker placement. Currently,
there are not any uniform guidelines regarding how to position a dog at the time of
marker placement, and the effect of limb position at the time of marker application
is unknown.
The purpose of the study reported here was to evaluate the effect of hindlimb position
at the time of marker application on sagittal plane kinematics of the hindlimb in
the dog. Our hypothesis was that changes in limb position at the time of marker application
would result in detectable differences in kinematic data obtained from dogs at a walk.
Materials and Methods
Animals
Six adult hound-breed dogs (three males and three females) from an established research
colony were evaluated in this study. All dogs were 2 to 3 years of age and weighed
20 to 30 kg. All dogs had normal bilateral hip and stifle radiographs, force plate
analysis, haematology, serum chemistry and physical examinations prior to study initiation.
The dogs were housed indoors in a climate-controlled environment and fed commercial
dog food ad libitum. Kinematic analysis protocol in this study was approved by the University of Georgia
Institutional Animal Care and Use Committee.
Spherical retroreflective markers (10 mm in diameter) were used to develop all kinematic models. All dogs were shorthaired and hair was not
clipped prior to marker application. Markers were affixed to the skin with double-sided
tape and cyanoacrylate. A three-dimensional testing space was established on a 13-m
walkway. Right-handed orthogonal coordinate axes were employed to describe the testing
space in three dimensions with 0,0,0 (X,Y,Z) located in the centre of the testing
space. Prior to each day’s collection, the system was calibrated with a calibration
frame of known dimensions (Vicon Peak Motus L-Frame, Vicon-Peak, Centennial, Colorado,
United States) and by dynamic linearization with a custom-made 0.700-m wand. Marker
locations were captured by a kinematic system of six infrared cameras (Vicon MX03,
Vicon Motion Systems, Inc., Centennial, Colorado, United States) arranged around the
gait platform. Data were recorded and analysed by a motion-analysis program (Peak
Motus 9.2; Vicon Motion Systems).
Right hindlimb kinematic data were collected from all dogs moving through the calibrated
space at a walk (velocity: 0.9–1.2 m/s;acceleration: ± 0.5m/s2). Truncal velocity and acceleration were recorded with a series of five photocells
placed 0.5 m apart and 0.5 m above the walkway. The first five valid trials were utilized
for analysis. All dogs were from an established research colony and as such were habituated
to the testing facility.
Kinematic Model
A sagittal plane kinematic model of the hindlimb in dogs was utilized in this study,
as previously described.[5],[10],[11] Five retroreflective markers (10 mm in diameter) were attached over anatomical landmarks
of the right hindlimb: the iliac crest, greater trochanter of the femur, the femorotibial
joint between the lateral epicondyle of the femur and the fibular head, the lateral
malleolus of the tibia and the distolateral aspect of the fifth metatarsal bone. All
skin markers were placed by the same experienced investigator (SK) throughout the study.
Data Collection
The kinematic models were established separately for each dog while standing in three
different positions: (1) the limb extended cranially with the foot cranial to the
pelvis (cranial), (2) the limb in a neutral position with the foot beneath the pelvis
(normal) and (3) the limb extended caudally with the foot caudal to the pelvis (caudal;[Fig. 1]). Kinematic data collection was performed separately for all limb positions (cranial,
normal and caudal) during the same testing period. The sagittal plane kinematic model
of the hindlimb was established in one of the standing positions, randomly assigned
for each dog prior to study initiation.[3] Following marker application, dynamic gait analysis was performed and five valid
kinematic trials were obtained for comparison. All markers were then removed with
acetone, leaving no visible trace. The investigator then positioned the dogs’ hindlimbs
in the second randomly selected position and repeated the above procedure. This was
performed in a similar manner for the third and final hindlimb position. Five valid
trials from each of the three initial standing positions were collected for each dog
and were used for kinematic analysis. A valid trial included a straight, forward walk
with no interruption or overt head movement. Throughout the study, all dogs were walked
by the same experienced handler.
Fig. 1 Figures and Tables
Data Analysis
Sagittal (flexion and extension) plane kinematics for the hip, stifle and tarsus were
generated and collected from each dog during a walk, separately for each of the three
initial standing positions. Complete waveform analysis was performed with Generalized
Indicator Function Analysis (GIFA) as previously described.[13],[14] All hypothesis tests were two sided and significance was set at p < 0.05. Kinematic variables, including joint range of motion and minimum and maximum
joint excursions were determined. Comparison between values obtained at different
initial standing positions was performed with one-way repeated measures analysis of
variance (ANOVA) with Greenhouse–Geisser correction. Post hoc pairwise comparison between the hindlimb positions was performed with a Bonferroni
adjustment. Normality was determined with the Shapiro–Wilk test. Statistical significance
was set at p < 0.05. The statistical analysis was performed with standard statistical software
(SPSS 16.0: Statistical Package for Social Science, Inc., Chicago, Illinois, United
States).
Results
Sagittal (flexion and extension) plane waveforms were generated for the hip, stifle
and tarsus for all three initial standing positions and were compiled graphically
with 95% confidence intervals ([Fig. 2]). The mean (±SD) values for range of motion and minimum/maximum joint angles of
the hip, stifle and tarsal joints were determined for all three hindlimb positions
(cranial, below, caudal) during marker application ([Table 1]).
Fig. 2 Graphs of mean extension and flexion angles (solid lines) of (A) hip, (B) stifle
and (C) tarsus with 95% confidence interval (dotted lines) for all dogs at walk after
marker placement in three hindlimb positions.
Table 1
Mean ± SD of maximum and minimum angles and range of motion of hip, stifle and tarsus
for all dogs at walk after marker placement in three hindlimb positions
|
|
Hip
|
Stifle
|
Tarsus
|
|
Max
|
Min
|
ROM
|
Max
|
Min
|
ROM
|
Max
|
Min
|
ROM
|
|
Cranial
|
140.0 ± 6.9[a]
|
102.3 ± 5.1[a]
|
37.7 ± 5.8
|
140.56 ±4.1[a]
|
105.36 ± 5.1[a]
|
35.26 ± 7.9
|
157.46 ± 5.2
|
119.76 ± 6.5
|
37.86 ± 6.6
|
|
Normal
|
147.6 ±4.2[a],[b]
|
111.1 ±4.4[b]
|
36.5 ± 4.0
|
148.06 ± 7.3[a]
|
113.86 ±4.3[b]
|
34.36 ± 7.3
|
158.36 ± 3.4
|
121.26 ± 6.6
|
37.16 ±4.7
|
|
Caudal
|
152.3 ± 3.3[b]
|
118.9 ± 3.2[c]
|
33.6 ±4.9
|
157.46 ± 7.2[b]
|
122.56 ± 8.1[c]
|
34.96 ± 5.5
|
160.06 ±4.8
|
124.36 ± 8.8
|
35.66 ± 5.7
|
|
p value
|
< 0.01
|
< 0.01
|
0.21
|
< 0.01
|
< 0.01
|
0.64
|
0.69
|
0.88
|
0.55
|
Abbreviations: Max, maximum angle; Min, minimum angle; ROM, range of motion.
Note: Values reported are degrees.
a Within the same column, values with different lowercase superscript letters differ
significantly (p < 0.05).
b Within the same column, values with different lowercase superscript letters differ
significantly (p < 0.05).
c Within the same column, values with different lowercase superscript letters differ
significantly (p < 0.05).
Each hindlimb positions generated grossly similar waveform shapes. However, when markers
were applied to the limb in the different standing positions, a vertical shift of
the waveform occurred. The direction of the shift was consistent for all joints with
the cranial standing position resulting in a downward shift and a caudal standing
position resulting in an upward shift of the waveform, as compared with the normal
standing position. The magnitude of this shift was least in the tarsus ([Fig. 2]).
Complete Waveform Analysis
GIFA found no significant difference between the waveforms from the hip and tarsus.
Significant differences were found between stifle waveforms obtained at different
initial standing positions.
Kinematic Variables
Significant differences were found in the maximum and minimum joint angles of the
hip and stifle between the different initial standing positions ([Table 1]). No differences were found between the minimum and maximum joint angles of the
tarsus. In all joints, the caudal standing position produced the largest angular values,
followed by the normal standing position, and then the cranial standing position produced
the smallest angular values. No differences were found in joint range of motion in
all joints for all three initial standing angles.
Discussion
The hypothesis in this study was supported. Different hindlimb positions at the time
of marker placement influenced sagittal joint angles of the hip and the stifle during
walking. Interestingly, the effect of the hindlimb positions on the tarsal joint was
negligible. While the wave for mshapes were visually similar for all hindlimb positions
at the time of marker placement, differences were detected between the stifle waveforms.
A vertical shift of the gait waveforms was found in the data obtained at different
initial standing positions. Subsequent to this shift, a change in the extreme values
of joint motion was noted. However, when the overall range of values was evaluated,
no differences were noted. This indicates that while differences in initial positioning
alter the specific values along the gait waveform, the average measure of motion throughout
the complete gait cycle is unaffected. Shifting of the waveforms along the vertical
axis has been documented in previous studies where changes in marker placement were
cited as the underlying cause.[1],[5],[12],[15] Marker placement over osseous anatomical landmarks is commonly performed to increase
consistency and reduce underlying soft tissue.[1],[5],[10] However, recent studies found that even when markers were placed over bone landmarks,
significant skin motion occurs.[11],[16] In those studies, the primary emphasis was on monitoring skin motion following marker
placement. In this study, the focus was instead on the effect of bone movement relative
to skin prior to marker application. Because the same investigator placed all markers
over easily palpable bone landmarks, the shifting of the waveforms is likely due to
a shifting of the bone underneath the skin during initial limb positioning at the
time of marker application.
Two-dimensional linkage kinematic models, as used in this study, are by design just
interconnected lines between estimated joint centres. Because of this, it is not surprising
that small changes in marker position resulted in changes to the measured joint angles–essentially
establishing a more acute or obtuse angular measurement. With the foot positioned
caudal to the pelvis, the hip and tarsus are extended (more obtuse angle) and with
the foot placed cranial to the pelvis, the hip and tarsus are flexed (more acute angle).
Consequently, skin markers placed while the hip was in extension shifted waveforms
of hip and stifle angles up along the y-axis, and markers placed during the hip flexion
shifted the waveforms down along the y-axis ([Fig. 2A, B]). In humans, it has been reported that with hip extension the greater trochanter
is located more cranially, and with hip flexion it is located more caudally.[17],[18] Interestingly, this is the first report in dogs to evaluate the effect of bone landmark
positioning, in relation to skin, at the time of marker application. Previous veterinary
studies have discussed and evaluated marker placement errors with no attention to
limb or bone positioning at the time of marker placement.[1],[2],[4],[5],[12] One such study on dogs examined stifle kinematics based on alternate greater trochanter
locations established with the dogs standing with identical limb positioning.[12] In that study, cranial deviation of the greater trochanter marker shifted the waveform
of the stifle angle up, and caudal deviation of the marker shifted the waveform down
the y-axis. Those findings as well as the current results in this study indicate that
it is not only marker application error that can lead to kinematic data variability,
but also inconsistent limb positioning during marker application. Therefore, consistent
limb positioning at the time of marker application is an important consideration in
kinematic studies.
Tarsal angles were not significantly affected by limb position during marker application.
The reason for these findings may be related to inherent differences in the marker
locations used to define tarsal motion (lateral femoral condyle, lateral malleolus
and distolateral aspect of the fifth metatarsal bone). Previous studies found that
while skin motion occurs in a cyclical pattern, there is a directional component of
motion that is marker dependent.[11],[16] In those studies, the lateral femoral condyle marker moved in a similar direction
to the long axis of the tibia, resulting in a change in the overall length of the
tibial segment.[11],[16] However, because of this unique directional change of the lateral femoral condyle
marker, skin motion results in a minimal overall change in the angular measurement
of the tarsus.[11] Furthermore, markers placed on the lateral malleolus and distolateral aspect of
the fifth meta-tarsal bone are in anatomical locations with minimal underlying soft
tissue–resulting in reduced skin motion artefact and marker placement error.
Marker placement error in kinematic analysis has been associated with a vertical shift
in the gait waveform, as seen in this study.[1],[5],[12],[15] In an effort to reduce the effect of this shift on kinematic analysis, normalization
procedures have been described.[5] However, the normalization process alters the original data and may mask true differences
between waveforms at different time points. Additionally, the normalization process
can be time consuming and therefore may not be well suited for clinical use. Various
methods have been introduced in dogs to avoid soft-tissue artefact or marker placement
error and to directly analyse bone movement.[9],[19],[20] Unfortunately, such methods are only applicable in a research setting due to invasiveness,
cost and speed of analysis. Regardless of clinical or research studies, the results
of this study show the importance of limb position at the time of marker application.
The use of a standardized limb position during marker application will reduce kinematic
data variability in both clinical and research studies.
Analysis of kinematic gait data has been performed with various methodologies. Historically,
Fourier analysis has been a common method used to compare joint angle waveforms from
dogs.[1],[4],[5],[15] In this study, GIFA was used to compare shapes of waveforms of joint angles as described
in recent studies.[12],[15],[21] These proved beneficial because unlike Fourier analysis, which is affected by the
vertical position of the gait waveforms, GIFA compares the true waveform shapes and
is unaffected by the position of the gait waveforms along the vertical axis.[13]–[15],[21] In addition to waveform analysis, comparison of peak joint angles and ranges of
motion was performed as previously described.[10],[11],[19],[22],[23] In this study, one-way repeated measures ANOVA allowed comparison between different
hindlimb positions within a subject and avoided the obscuring caused by averaging
all values, as has been seen in previous studies using only Fourier analysis.[5],[7]
Numerous sources of variability in kinematic gait analysis have been identified. Recent
research has documented that there are inter-examiner variability and intra-examiner
variability between testing days in the kinematic analysis of the stifle.[15] To avoid inter-examiner variability in the current study, all markers were applied
by a single experienced examiner. To reduce intra-examiner variability in marker placement
between days as well as to avoid true gait changes between days, all kinematic data
were collected during the same testing period for individual dogs.[24] Also, the same dog handler walked all dogs in a narrow range of velocity to prevent
possible interference with gait attributable to behaviour response to the handler.
Limitations of this study include the small number of dogs. Additionally, while the
results of this study and previous studies indicate that different hindlimb positions
contribute to the variation of the skin marker placement, this study did not measure
actual differences in individual marker locations resulting from different hindlimb
positions. This study still showed how kinematic data are affected by limb position
during marker placement. Further research is warranted to characterize skin marker
placement during various body postures.
Overall, the results of this study show that sagittal plane kinematic data were affected
by hindlimb position at the time of marker application. Additionally, with the kinematic
model utilized in this study the direction of shift in the gait waveforms for all
joints was consistent for the cranial and caudal limb positions. Consistent, standardized
positioning during marker placement is important to reduce variability necessary for
clinical gait analysis such as kinematics and inverse dynamics.
