Effect of trotting speed on kinematic variables

AJVR—16-11-0301R—Cruz—1fig—4tab—VLS—CAS

Effect of trotting speed on kinematic variables measured by use of extremity-mounted inertial measurement units in nonlame horses performing controlled treadmill exercise OBJECTIVE To assess effects of speed on kinematic variables by use of extremitymounted inertial measurement units (IMUs) in nonlame horses performing controlled exercise on a treadmill.

Antonio M. Cruz dvm, mvm, phd Beatriz Vidondo msc biol, msc it, phd Alessandra A. Ramseyer dr med vet Ugo E. Maninchedda dr med vet, msc

ANIMALS 10 nonlame horses.

Received November 11, 2016. Accepted May 9, 2017. From the Institute Suisse de Médicine Équine, Vetsuisse-Fakultät, Universität Bern, Bern 3012, Authors: Switzerland (Cruz, Ramseyer, Maninchedda); and the Please verify Veterinary Public Health Institute, Vetsuisse-Fakultät, Universität Bern, Bern 3097, Switzerland (Vidondo). the underDr. Cruz’s present address is EU-Universidad Cardenal lined postal Herrera, Facultad de Veterinaria, Departamento de code is cor- Medicina y Cirugía Animal, C/ Tirant lo Blanc 7, 46115 rect. Alfara del Patriarca-Valencia, Spain. Address correspondence to Dr. Cruz (acruz4@mac. com).

PROCEDURES 6 IMUs were attached at predetermined locations on 10 nonlame Franches Montagnes horses. Data were collected in triplicate during trotting at 3.33 and 3.88 m/s on a high-speed treadmill. Thirty-three selected kinematic variables were analyzed. Repeated-measures ANOVA was used to assess the effect of speed. RESULTS Significant differences between the 2 speeds were detected for most temporal (11/14) and spatial (12/19) variables. The observed spatial and temporal changes would translate into a gait for the higher speed characterized by an increased stride length, protraction and retraction, flexion and extension, mediolateral movement of the tibia, and symmetry, but with similar temporal variables and a reduction in stride duration. However, even though the tibia coronal range of motion was significantly different between speeds, there was large variability for this variable, which raised concerns about whether these changes were clinically relevant. For some variables, the lower trotting speed apparently was associated with more variability than was the higher trotting speed. CONCLUSIONS AND CLINICAL RELEVANCE At a higher trotting speed, horses moved in the same manner (eg, the temporal events investigated occurred at the same relative time within the stride). However, from a spatial perspective, horses moved with larger action of the segments evaluated. The detected changes in kinematic variables indicated that trotting speed should be controlled or kept constant during gait evaluation. (Am J Vet Res 2018;79:211–218)

O

ne of the main tasks of equine veterinarians in clinical practice is to perform gait assessment of horses, whether for sport selection or as part of prepurchase examinations, ongoing performance evaluations, or lameness investigations. To perform this task, veterinarians primarily rely on direct visual evaluation of various aspects such as symmetry of movement, harmony, and cadence during walking, trotting, and cantering. For trotting horses, lack of symmetry during the phases of the stride (manifested as head nodding, pelvic asymmetry, or uneven extension of the metacarpophalangeal joints) is the primary indicator used by veterinarians to detect lameness.1 However, direct visual evaluation can be ABBREVIATIONS IMU Inertial measurement unit ROM Range of motion

subjective and therefore an inaccurate estimator of gait changes.2 To eliminate subjectivity associated with traditional gait assessment, efforts have been made to develop portable kinematic gait analysis methods that could be used in field conditions. One of these methods involves the use of inertial technology by instrumenting a horse’s extremities with IMUs to record selected spatial and temporal variables.3 The reliability for the coefficients of repeatability of such a system have recently been confirmed.3 Development of miniaturized IMUs has facilitated measurements of horses in nonlaboratory settings (ie, real-life conditions).4 Use of trunk- and extremity-mounted IMUs has been investigated in equine locomotion studies.4–7,a They have the potential for easy and practical clinical application because they enable simultaneous subjective evaluation by an examiner and collection and

AJVR • Vol 79 • No. 2 • February 2018

211

analysis of objective data while a horse is trotting. The sensors are small, robust, and lightweight; thus, they do not have a major impact on the movement of a horse.4 Several studies5,6,8 in which some aspects of IMUs were compared with optoelectronic systems in horses have revealed good correlation between both systems. Prior to clinical implementation of an IMU system, it would seem appropriate to investigate the effects that various measurement conditions may have on the data output; therefore, there was a need for the study reported here. During examination of a horse, one of the variables that plays an important role is the speed at which a horse is evaluated because speed affects kinematic variables and reduces the ability to detect gait events by use of conventional means.9–12 The effect of speed on gait kinematic variables has been defined by use of analysis with optical means under laboratory conditions.12 However, the effect of speed on kinematic variables has not been evaluated by the use of extremity-mounted IMUs; therefore, defining changes in selected kinematic variables associated with speed would provide additional information toward the use of IMUs or gait assessment (ie, lameness evaluation) in a clinical setting. Investigators of some previous studies9,11–16 have evaluated the effect of speed on temporal and spatial variables (eg, stride length, stride frequency, swing duration, stance duration, restraint phase, propulsion phase, and diagonal dissociation). However, the effects of speed on other temporal and spatial variables (eg, limb phasing, symmetry of the distal aspect of a limb, and sagittal and coronal ROMs) that can be measured with IMUs have not been described. The objective of the study reported here was to evaluate the effect of 2 trotting speeds on selected kinematic variables and to determine the effect of trotting speed on some previously unreported variables.

Materials and Methods Animals

Ten horses were arbitrarily selected from a herd of Franches Montagne stallions at the Haras National Stud Center in Avenches, Switzerland. These horses had been used in a previous study3 conducted to investigate the coefficients of repeatability for the extremity-mounted IMUs. Selected horses ranged from 6 to 19 years of age, were in good physical condition, and had good conformation. Horses were excluded if they were lame, unhealthy, or receiving medication. Mean ± SD body weight of the horses was 537.5 ± 28.3 kg, and mean height (from the ground to the highest point of the shoulders [withers]) was 157.4 ± 1.4 cm. All horses were accustomed to exercise on a treadmillb in accordance with standard procedures.17 Informed consent was obtained for use of the horses in the study; the study was approved by the Swiss Vaud cantonal authority (VD protocol 3041) and was conducted in accordance with institutional guidelines for humane animal treatment. 212

To confirm similarity among size of limb segments, other variables for each horse were recorded. These included length of the left and right radii, third metacarpal bones, third metatarsal bones, and tibias by use of a flexible measuring tape and identification of palpable landmarks (Figure 1). Length of a radius was the distance from the lateral tuberosity of the proximal portion of the radius to the styloid process (which corresponded to the lateral styloid process of the ulna). Length of a third metacarpal bone was the distance from the proximal extremity to the palpable condylar fossa; length of a third metatarsal bone was measured by use of these same landmarks. Length of a tibia was the distance from the lateral condyle to the malleolus lateralis (which corresponded to the lateral malleolus of the fibula).

Study design

A randomized prospective study was performed during a time span of 2 weeks. Trotting speeds were 3.33 and 3.88 m/s. At each session, data were collected for both speeds during a 15-minute period, with the slow speed always preceding the fast speed.

IMU system

Data collection was performed with commercially available IMUs.c The system included software, 6 sensors, and a laptop computer.18 Technical specifications for the IMUs have been described in another study.3 Each IMU was mounted into a brushing bootd and tibia straps. Samples were collected with a 12-bit analog-to-digital converter at a frequency of 102.4 Hz. The IMUs were time stamped and synchronized at the start of each horse’s trials by a pulse sent simultaneously to the IMUs by use of specifically written software.e This software was also used for recording and automated processing of the data. Sensors contained a precision clock and memory storage service card. Sensors were programmed to determine the cycle associated with each stride via a proprietary algorithm. The metacarpus sagittal ROM outputs produced by this system have been compared to results for an optical system.5 In that study,5 precision bias was reported, and it revealed the ability of the IMU system to measure subtle changes in stride phases with no bias, precision of 0.025%, and ROM with a mean ± SD bias of 1.6 ± 1.9°. The authors are aware of no publications that support the accuracy of the IMU system when measuring segment displacement,5–7 angular ROM,19 and stride frequency.9,10 In human medicine, IMU systems have been established as acceptable and reliable,20,21 and repeatability of the IMU system measurements for horses during controlled conditions of treadmill exercise has also been evaluated and was found to be good.3

IMU variables

The IMU system used in the study was capable of simultaneously capturing the entire motion cycle of the metacarpal bones, metatarsal bones, and tibias

AJVR • Vol 79 • No. 2 • February 2018

Figure 1—Diagrams of a radius (A), tibia (B), and metacarpal or metatarsal bones (C) of a horse depicting the landmarks used for measuring bone length (dotted line). Length of a radius was the distance from the lateral tuberosity of the proximal portion of the radius to the styloid process (which corresponded to the lateral styloid process of the ulna). Length of a third metatarsal bone was the distance from the proximal extremity to the palpable condylar fossa; length of a third metacarpal bone was measured by use of these same landmarks. Length of a tibia was the distance from the lateral condyle to the malleolus lateralis (which corresponded to the lateral malleolus of the fibula).

of all limbs. The spatial-temporal variables recorded by the system defined some aspects of the kinematic characteristics during trotting. The IMU system recorded 14 temporal gait variables. This included the temporal phase-lag between respective limbs, also known as limb-phasing variables. Phasing is defined through a cross-correlation approach of the rotation velocity around the lateromedial axis of the inertial sensor on a stride-by-stride basis. Therefore, phase-lag for each limb is expressed as a percentage of the stride duration of a reference limb.22 Stride duration and timing for maximal protraction and retraction of third metacarpal and third metatarsal bones (expressed as the percentage of the stride) were also recorded. Diagonal asymmetry is the difference between the diagonal limb-phasing timing couplets and was calculated by use of the following equation: diagonal asymmetry (%) = (LF – RH) – (RF – LH), where LF, RH, RF, and LH are results for the left forelimb, right hind limb, right forelimb, and left hind limb, respectively, and the value for LH was 0% because it was the reference limb. The IMU system recorded 19 spatial gait variables. The ROM (in degrees) referred to the sagittal

angles of the carpi and tarsi joints and segment angles of the third metacarpal bones, third metatarsal bones, and tibias. Coronal movement (in degrees) referred to the metacarpal bones, metatarsal bones, and tibias. Symmetry of the ROM for each segment was calculated as the value for the left side minus the value for the right side divided by the mean value for both sides. Except for the angles of the tarsal joint, the rest of the angles, which defined the ROM in the study, were sagittal and coronal segment angles. A sagittal segment angle is the angle that results when the segment subtends from its maximum retracted position to its maximum protracted position. A coronal angle is the maximum range the segment moves through in the frontal plane during the stride. A negative coronal ROM indicates medial movement or adduction, and a positive ROM indicates lateral movement or abduction. The tarsal angle is the angle subtended between 2 segments, the tibia and third metatarsal bone. Another spatial variable that was calculated was stride length.

AJVR • Vol 79 • No. 2 • February 2018

213

Data collection

Treadmill speed was calibrated by use of a magnetic speedometer system f attached to a racing bicycle by conducting 3 consecutive trials that had identical results between the treadmill settings and speedometer readings. Prior to the beginning of each data collection session, all horses were allowed to warm up by hand walking for 20 minutes. Exercise was performed on the treadmill by walking horses for 10 minutes and then trotting them for 5 minutes to reach steady-state locomotion.17 Horses were placed in a stationary position on the treadmill, and the IMUs were synchronized and time stamped. Brushing boots (outside length, 23.5 cm; inside length, 14.5 cm) were mounted on the third metacarpal and third metatarsal bones of each limb of the horses, and custom-made elastic straps (width, 5 cm) were attached to the groove just dorsal to the gastrocnemius tendon. Boots and straps were provided with hook-and-loop fasteners. A small custom-fitted pouch on the lateral aspect of each boot (15 cm proximal to the metacarpophalangeal joints) and strap (10 cm proximal to the tarsal joints) was designed to hold the sensor firmly to reduce motion and to facilitate synchronicity with limb movement throughout the data collection period. After the straps and boots were placed, each

horse was walked and trotted on the treadmill until the gait appeared visually normal (usually attained within 2 minutes). The treadmill was then stopped, and the 6 sensors were turned on by clicking the sensor switch immediately before placing them in their specified locations. The sensors were all visually aligned to the vertical axis of the limb segment in the lateral aspect of the limb. After all sensors were placed and turned on, the horse remained still for 10 seconds to help the sensors find a stationary period to self-calibrate (a minimum stationary period of 10 seconds was a prerequisite for sensor calibration). This calibration enabled the system to define the gravitational vector. The same handler held all horses on the left side of each horse by use of a loosely hanging rope. All horses were then walked at a constant speed of 1.8 m/s until data were collected during trotting for 30 strides at 3.33 m/s and then during trotting for 30 strides at 3.88 m/s. This process was repeated 3 times, with a 10-second interval between successive trials. Data were collected twice with a 1-week interval between sessions. External factors (eg, noise or moving objects) that could have influenced measurement results were eliminated whenever possible. If a horse’s degree of distraction or excitement was definitively noticeable, the measurement trial was discarded and repeated

Table 1—Mean ± SD values for selected temporal gait variables measured by use of extremitymounted IMUs for 10 nonlame horses trotting at 2 speeds.

Trotting speed (m/s)

Variable

Difference

3.3

3.8

Mean

%

P value*

0.73 ± 0.02 2.42 ± 0.08

0.69 ± 0.02 2.55 ± 0.16

0.04 0.13

5.47 5.37

< 0.001 < 0.001

63.05 ± 1.48 13.24 ± 1.52 49.40 ± 0.75

63.77 ± 1.42 14.10 ± 1.58 49.49 ± 0.89

0.72 0.86 0.09

1.14 6.49 0.18

< 0.001 < 0.001 0.4

Protraction (%)† Timing of maximal metatarsal protraction Left 47.70 ± 2.35 Right 48.70 ± 2.73

48.73 ± 2.63 49.63 ± 2.46

1.03 0.93

2.15 1.90

0.001 0.004

Timing of maximal metacarpal protraction Left 53.13 ± 2.22 Right 52.73 ± 2.39

55.13 ± 2.42 54.87 ± 2.48

2.00 2.14

3.76 4.05

< 0.001 < 0.001

Stride duration (s) Stride length (m) Limb phasing (%) Left forelimb Right forelimb Right hind limb

Retraction (%)† Timing of maximal metatarsal retraction Left 4.48 ± 2.62 Right 5.16 ± 2.49 Timing of maximal metacarpal retraction Left 18.80 ± 1.53 Right 18.20 ± 2.07

4.10 ± 1.92 4.48 ± 2.12

0.38 0.68

8.48 13.14

0.72 0.53

18.00 ± 1.33 17.27 ± 1.84

0.80 0.93

4.25 5.10

< 0.001 < 0.001

Symmetry (%)† Diagonal‡

–0.18 ± 1.33

0.23

56.09

< 0.001

–0.41 ± 1.08

*Values were considered significant at P < 0.05. †Represents relative values as a percentage of the stride as referenced to events in the left hind limb (which was assigned a value of 0%). ‡Values were calculated as follows: Diagonal asymmetry (%) = (LF – RH) – (RF – LH), where LF, RH, RF, and LH are results for the left forelimb, right hind limb, right forelimb, and left hind limb, respectively, and the value for LH is 0% because it is the reference limb.

214

AJVR • Vol 79 • No. 2 • February 2018

Table 2—Mean ± SD values for selected spatial gait variables measured by use of extremitymounted IMUs for 10 nonlame horses trotting at 2 speeds.

Trotting speed (m/s)

Variable

Difference

3.3

3.8

Mean

%

P value*

Sagittal ROM (°) Left tarsus Right tarsus Left tibia Right tibia Left metacarpus Right metacarpus Left metatarsus Right metatarsus

41.58 ± 3.48 37.14 ± 3.96 45.62 ± 3.45 44.24 ± 4.08 84.39 ± 2.35 85.03 ± 4.10 53.42 ± 2.89 52.53 ± 4.69

44.08 ± 3.81 40.12 ± 3.40 50.69 ± 3.46 48.06 ± 3.91 91.14 ± 2.55 91.43 ± 3.73 58.57 ± 2.64 58.33 ± 3.09

2.50 2.98 5.07 3.82 6.75 6.40 5.15 5.80

6.01 8.02 11.11 8.63 7.99 7.52 9.64 11.04

< 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001

Coronal ROM (°) Left tibia Right tibia Left metacarpus Right metacarpus Left metatarsus Right metatarsus

21.39 ± 3.70 21.56 ± 3.23 13.24 ± 4.38 17.07 ± 7.46 15.70 ± 5.67 14.20 ± 5.23

22.66 ± 3.76 23.20 ± 3.33 13.92 ± 4.85 16.94 ± 8.29 16.88 ± 6.43 15.38 ± 5.76

1.27 1.64 0.68 0.13 1.18 1.18

5.93 7.60 5.13 0.76 7.51 8.30

0.004 < 0.001 0.21 0.90 0.19 0.11

Symmetry (%)† Hind asymmetry Tarsus symmetry Tibia symmetry Metacarpus symmetry Metatarsus symmetry

0.60 ± 0.75 11.46 ± 12.96 3.20 ± 9.98 –0.68 ± 3.45 2.00 ±10.09

0.51 ± 0.89 9.35 ± 11.48 5.43 ± 9.43 –0.27 ± 3.65 0.44 ± 4.11

0.09 2.11 2.23 0.41 1.56

15.00 18.41 69.68 60.29 78.00

< 0.001 0.23 0.04 0.37 0.78

See Table 1 for key.

again immediately, assuming the conditions then were optimal.

IMU data analysis

During the data collection sessions, information was stored on the sensors. At the end of each horse’s data collection session, the sensors were removed from the pockets and immediately switched off. Data then were downloaded from the sensors onto a personal computer via a USB connector. Downloaded data were analyzed with the software by use of a proprietary algorithm18 to produce and display processed data for the accelerometer and gyroscope signal output in the form of orientation and temporal data. The authors manually and visually selected a window of at least 10 strides with steady locomotion for use in the analysis. For the purpose of this study, steady locomotion was characterized by a regular signal from each sensor and regular stride duration, as could be seen on the output screen for the recording period. The orientation and temporal events of each segment were used by the system to calculate the joint angles as a relationship of 2 adjacent segments and limb phasing as the relative timing between segments.18 The preselected kinematic variables that the system generated defined the spatial orientation of each limb, and temporal data defined the relative intralimb and interlimb movements. Stride duration calculated from the software was used to determine stride length over the region chosen for the gait analysis by use of the following equation: stride length = trotting speed X stride duration.

Statistical analysis

Statistical analysis of data was conducted with commercial software.g Data were examined for the presence of outliers. Values less than the 2.5th and greater than the 97.5th percentiles were considered outliers. All variables were tested for normality (Shapiro-Wilk test), and a repeated-measures ANOVA for each of the gait analysis variables was performed with horse as the subject variable and measurement repetition (1 to 3) and speed (1 = slow trotting speed and 2 = fast trotting speed) as fixed factors. To make comparisons with results of a repeatability study,3 repeatability coefficients at each of the 2 trotting speeds were calculated as described elsewhere23 by use of the following equation: 1.96 X √2 X withinsubject SD. In turn, within-subject SD was obtained by computation of a 1-way ANOVA h for each outcome variable with respect to the variable horse. An estimated repeatability coefficient of 1 means that the absolute difference between any 2 future measurements made on a particular horse is estimated to be ≤ 1 on 95% of occasions.23 For all analyses, values were considered significant at P < 0.05.

Results Horses

Mean ± SD radius length was 38.14 ± 3.99 cm, and mean tibia length was 38.28 ± 1.14 cm. Mean length of the third metacarpal bone was 25.35 ± 0.62 cm, and mean length of the third metatarsal bone was 30.57 ± 0.78 cm. All data were normally distributed.

AJVR • Vol 79 • No. 2 • February 2018

215

Temporal variables

Significant differences were associated with trotting speed for most (11/14) temporal variables. Trotting at 3.88 m/s resulted in a significant (P < 0.001) reduction in stride duration (5.47%) and a significant (P < 0.001) increase in stride length (5.37%), compared with results when trotting at 3.33 m/s (Table 1).

Spatial variables

Significant differences were associated with trotting speed for 12 of 19 spatial variables (Table 2). All of the sagittal ROM values increased significantly (P < 0.001) when trotting at 3.88 m/s, compared with results when trotting at 3.33 m/s, and this range of variation fluctuated from 5% to 11% in all segments, whereas the coronal ROM was significantly different for only the left (P = 0.004) and right (P < 0.001) tibia, which had an increase of 5.6% and 7.0%, respectively, associated with speed. Repeatability coefficients of the temporal variables ranged from 0.10 to 8.74 and were similar to previously reported values3 (Table 3). Variability of the spatial variables (specifically the coronal ROM and symmetry variables) was slightly higher than previously reported values (Table 4). However, this was partly attributable to 2 outliers for the slower trotting speed and 1 outlier for the higher trotting speed. ReTable 3—Repeatability coefficients* for selected temporal gait variables measured by use of extremity-mounted IMUs for 10 nonlame horses trotting at 2 speeds. Variable

Discussion Results of the study reported here indicated that most analyzed spatial and temporal gait variables changed with trotting velocity. The findings for stride duration and stride length are in agreement with results of other studies.11,12 Specifically, there was an adaptation to an increase in speed by increasing stride length and reducing stride duration (eg, increased stride frequency). This increase in frequency accelerated the events during movement, which may make them harder to detect visually, and this further supported the need to develop an objective method of gait analysis, particularly at higher speeds because technological analysis of the gait is more sensitive than direct visual assessments.24 To our knowledge, the study reported here was the first in which extremity-mounted IMUs were used to provide results for output variables (eg, limb phasing and coronal ROM of the radius and tibia) that have not been reported previously. All variables related to sagittal ROM significantly increased in association with trotting speed. As the speed increased in the present study, the extremities appeared to have a 6% to 11.5% change in ROM as-

Table 4—Repeatability coefficients* for selected spatial gait variables measured by use of extremity-mounted IMUs for 10 3.3 3.8 nonlame horses trotting at 2 speeds. Trotting speed (m/s)

Limb phasing (%) Stride duration (s) 0.03 0.02   Stride length (m) 0.10 0.28 Left forelimb 2.82 2.26 Right forelimb 2.66 2.39 Right hind limb 1.74 1.86 Protraction (%) Timing of maximal metatarsal protraction Left 3.93 5.14 Right 5.03 4.95 Timing of maximal metacarpal protraction Left 4.12 3.81 Right 3.30 3.33 Retraction (%) Timing of maximal metatarsal retraction Left 8.50 4.22 Right 8.74 5.14 Timing of maximal metacarpal retraction Left 3.79 3.07 Right 3.62 3.07 Symmetry (%) Diagonal asymmetry 2.30 2.84 *Repeatability coefficients were calculated as 1.96 X √2 X withinsubject SD, as described elsewhere.23 In turn, within-subject SD was obtained by computing a 1-way ANOVA for each outcome variable with respect to the variable horse. An estimated repeatability coefficient of 1 means that the absolute difference between any 2 future measurements made on a particular horse is estimated to be ≤ 1 on 95% of occasions.23

216

moval of these outlier data points for certain variables lowered the repeatability coefficients to previously published values,3 except for symmetry of the tarsus, for which the coefficients of repeatability were approximately 26 to 28 for both trotting speeds, compared with the previously reported value of 16 to 18.3

Variable Sagittal ROM (°) Left tarsus left Right tarsus Left tibia Right tibia Left metacarpus Right metacarpus Left metatarsus Right metatarsus Left tibia ROM (°) Right tibia ROM (°) Coronal ROM (°) Left metacarpus Right metacarpus Left metatarsus Right metatarsus Symmetry (%) Tarsus symmetry Tibia symmetry Metacarpus symmetry Metatarsus symmetry Hind asymmetry

Trotting speed (m/s) 3.3 3.8 8.60 8.02 4.99 8.96 4.35 5.41 6.37 12.25 (6.41)†

9.06 7.65 5.71 6.26 5.38 5.98 6.90 6.65

7.06 6.43

6.57 6.83

8.37 15.58 13.97 10.55

8.60 17.54 14.45 12.07

30.74 (28.54)† 20.29 (13.28)† 6.23 28.84 (10.43)† 1.74

27.15 (26.49)† 13.02 7.78 9.88 1.86

†Values in parentheses are the recalculated repeatability coefficients after removal of data points that were outliers (2 for the slower trotting speed and 1 for the higher trotting speed). See Table 3 for remainder of key.

AJVR • Vol 79 • No. 2 • February 2018

sociated with speed. Other spatiotemporal changes associated with speed may also serve as reference values when using an IMU system. However, linear variations associated with trotting speed could not be evaluated because only 2 trotting speeds were used. In the coronal plane, only movement of both tibias had a significant increase associated with trotting speed. This increase in coronal ROM is coupled with flexion-extension of the tarsus joint at a rate of approximately 1° of abduction for each 3° of flexion,25 which is the approximate difference also detected in the study reported here. However, even though there were significant differences in the tibia coronal ROM, there was a lot of variability, which raised concerns as to whether these changes for this variable would be clinically relevant. The detected spatial and temporal changes would translate into a gait characterized by a longer stride length, more protraction-retraction, more flexion-extension, and slightly more symmetry in the movements, while maintaining similar temporal variables. Thus, from a temporal variable perspective, a horse trotting at a higher speed would move in the same manner (eg, temporal events would occur at the same relative time within the stride), but from a spatial perspective, a horse would move with larger ROM of the segments. Because measurements provided by the IMU system were often not absolute values, the detected changes must be interpreted in light of the measured stride duration of 690 to 730 milliseconds. For example, limb phasing of the forelimbs changed with trotting speed and decreased within a magnitude of 4 to 6 milliseconds. For the protraction and retraction variables, the maximum change detected was 2.14%, which was equivalent to 13.2 milliseconds of real time. These changes were within the SD of 20 milliseconds measured for stride duration and likely within the range of typical biological variability, which probably indicated that limb phasing and protraction and retraction variables will remain constant despite alterations in trotting speed. Regarding applicability of the IMU system, results for the present study were similar to previously reported results for some variables. For example, tarsal sagittal ROM has been recorded by means of optical systems.25,26 In those studies, mean ± SD tarsus ROM at a trotting speed of 4 m/s was 55.4 ± 7.6° (left tarsus) and 51.5 ± 4.1° (right tarsus), which differs slightly from the values for the study reported here when horses trotted at 3.8 m/s (44.08 ± 3.8° and 40.12 ± 3.40° for the left and right tarsus, respectively). These discrepancies could be attributed to methodological differences (eg, velocity and breed [warmbloods vs Franches Montagne]) between studies that involved the use of an IMU or optical system. Measurements for optical systems are based on the presence of skin markers at predetermined locations, which are subject to error attributable to marker displacement,27 whereas measurements for IMUs involve processing of the accelerometer and gyroscope signals to gener

ate orientation and temporal data. The IMU system calculates the orientation and temporal events of each segment; it then calculates the joint angles as a relationship of 2 adjacent segments and the limb phasing as the relative timing between segments.18 The present study had certain limitations. We chose to exercise the horses on a treadmill in an effort to control all other elements of variability to ensure that any changes were attributable to trotting speed and not to other factors (eg, environmental or behavioral). Treadmill locomotion differs from overground locomotion, and it was possible that the changes detected in the present study would not be applicable to field situations. However, we believe that although the absolute effects may differ between treadmill and overground locomotion, the relative changes and patterns detected in the study reported here will be maintained for field situations because they appear to be logical on the basis of our clinical experiences. We are aware of no information pertaining to the speed at which horses usually are trotted during gait evaluation. On the basis of our clinical experience, horses are trotted at a naturally selected speed at which the horse appears most comfortable. Empirically, this is done by leading a horse during the examination, leaving the head free, and avoiding any interference and irregular movements. Therefore, for the study reported here, trotting speeds were selected on the basis of our experience with previous experiments that involved use of this population of horses and selection of minimal and maximal speeds at which all horses appeared to be comfortable while trotting on the treadmill and that would not cause the horses to break into an irregular gait. External factors were controlled as much as possible, in particular in the selection of a homogenous horse population (as evident from the signalment and size descriptors) with similar training programs. This avoided the need to perform scaling procedures. Therefore, we believe that the differences detected in kinematic variables were mostly attributable to the direct effect of trotting speed. Effects for each individual horse cannot be ruled out, but we selected the number of horses to provide us with sufficient statistical power, as determined on the basis of a previous study3 conducted by our research group. Because outliers were detected, we reiterate, as stated for a previous study,3 the need to obtain at least 3 measurements (or even better, 5 measurements) and to carefully select a window of strides during the software analysis that does not contain sudden changes in stride duration. The relatively low repeatability of the symmetry of the tarsus for both trotting speeds was somewhat unexpected. Future studies should be designed to discern whether this is a characteristic of this key joint of the equine anatomy or was an effect attributable to the group of horses used in the present study. Because the present study was performed with a homogenous group of horses in controlled conditions, additional studies that involve

AJVR • Vol 79 • No. 2 • February 2018

217

horses of other breeds and sizes and different conditions would seem relevant. For the study reported here, the IMU system could be used to detect kinematic changes associated with trotting speed during controlled treadmill exercise that were in agreement with values previously reported for some temporal and spatial variables. Other differences associated with trotting speed have not been previously reported and constituted new information. Controlling trotting speed or maintaining a constant trotting speed was important to minimize variations in a horse’s movements.

Acknowledgments Supported by the Institute Suisse de Médicine Équine. The authors thank Marie Mayerat and Christian Herren for assistance with handling of the horses and Joëlle Stutz for providing the images.

Footnotes a.

b. c. d. e.

f. g. h.

Cruz AM, Dieckmann B, van Iwaarden S, et al. Inertial motion units to assess gait kinematic variables in healthy Icelandic horses (abstr), in Proceedings. 25th Annu Sci Meet Eur Coll Vet Surg 2016;73. Kagra, Graber AG, Fahrwangen, Switzerland. Pegasus GaitSmart, provided by European Technology for Business, Codicote, Hertfordshire, England. Woof Wear, Bodmin, Cornwall, England. Poseidon, version 4.0, provided by European Technology for Business, Codicote, Hertfordshire, England. Anima+, TwoNav, Arenys de Mar, Spain. NCSS, version 10, NCSS LLC, Kaysville, Utah. Function oneway, Stata, version 14, StataCorp, College Station, Tex.

References 1.

Stashak TS, Baxter GM. Examination for lameness. In: Baxter GM, ed. Adams and Stashak’s lameness in horses. 6th ed. Philadelphia: Lippincott Williams & Wilkins, 2011;109–205. 2. Keegan KG. Evidence-based lameness detection and quantification. Vet Clin North Am Equine Pract 2007;23:403–423. 3. Cruz AM, Maninchedda UE, Burger D, et al. Repeatability of gait pattern variables measured by use of extremity-mounted inertial measurement units in nonlame horses during trotting. Am J Vet Res 2017;78:1011–1018. 4. Pfau T. Practical assessment of movement symmetry in horses: use of inertial sensors during clinical lameness exams. Available at: www.befred.org/docs/2423.pdf. Accessed Sep 8, 2015. 5. Roepstorff L, Wiestner T, Weishaupt MA, et al. Comparison of microgyro-based measurements of equine metatarsal/ metacarpal bone to a high speed video locomotion analysis system during treadmill locomotion. Vet J 2013;198:e157– e160. 6. Olsen E, Andersen PH, Pfau T. Accuracy and precision of equine gait event detection during walking with limb and trunk mounted inertial sensors. Sensors (Basel) 2012;12:8145–8156. 7. Starke SD, Witte TH, May SA, et al. Accuracy and precision of hind limb foot contact timings of horses determined using a pelvis-mounted inertial measurement unit. J Biomech 2012;45:1522–1528.

218

8.

Nankervis K, Hodgins D, Marlin D. Comparison between a sensor (3D accelerometer) and ProReflex motion capture systems to measure stride frequency of horses on a treadmill. Comp Exerc Physiol 2008;5:107–109. 9. Galisteo AM, Cano MR, Morales JL, et al. The influence of speed and height at the withers on the kinematics of sound horses at the hand-led trot. Vet Res Commun 1998;22:415– 423. 10. Starke SD, Raistrick KJ, May SA, et al. The effect of trotting speed on the evaluation of subtle lameness in horses. Vet J 2013;197:245–252. 11. Leach DH, Drevemo S. Velocity-dependent changes in stride frequency and length of trotters on a treadmill. In: Equine exercise physiology. Vol 3. Davis, Calif: ICEEP Publications, 1991;136–140. 12. van Weeren PR, van den Bogert AJ, Back W, et al. Kinematics of the Standardbred trotter measured at 6, 7, 8 and 9 m/s on a treadmill, before and after 5 months of prerace training. Acta Anat (Basel) 1993;146:154–161. 13. Clayton HM. Comparison of the stride kinematics of the collected, working, medium and extended trot in horses. Equine Vet J 1994;26:230–234. 14. McLaughlin RM Jr, Gaughan EM, Roush JK, et al. Effects of subject velocity on ground reaction force measurements and stance times in clinically normal horses at the walk and trot. Am J Vet Res 1996;57:7–11. 15. Barr AR, Dow SM, Goodship AE. Parameters of forelimb ground reaction force in 48 normal ponies. Vet Rec 1995;136:283–286. 16. Weishaupt MA, Hogg HP, Auer JA, et al. Velocity-dependent changes of time, force and spatial parameters in Warmblood horses walking and trotting on a treadmill. Equine Vet J Suppl 2010;42:530–537. 17. Buchner HH, Savelberg HH, Schamhardt HC, et al. Habituation of horses to treadmill locomotion. Equine Vet J Suppl 1994;17:13–15. 18. Whalley A, Hodgins D. Assessment of gait. US patent 8,715,208 B2. May 6, 2014. 19. Cooper G, Sheret I, McMillan L, et al. Inertial-sensor based knee flexion/extension angle estimation. J Biomech 2009;42:2678–2685. 20. Cuesta-Vargas AI, Galán-Mercant A, Williams JM. The use of inertial sensors system for human motion analysis. Phys Ther Rev 2010;15:462–473. 21. Monda M, Goldberg A, Smitham P, et al. Use of inertial measurement units to assess age-related changes in gait kinematics in an active population. J Aging Phys Act 2015;23:18–23. 22. Hildebrand M. Symmetrical gaits of horses. Science 1965;150:701–708. 23. Bartlett JW, Frost C. Reliability, repeatability and reproducibility: analysis of measurement errors in continuous variables. Ultrasound Obstet Gynecol 2008;31:466–475. 24. McCracken MJ, Kramer KG, Keegan KG, et al. Comparison of an inertial sensor system of lameness quantification with subjective lameness evaluation. Equine Vet J 2012;44:652– 656. 25. Lanovaz JL, Khumsap S, Clayton HM, et al. Three-dimensional kinematics of the tarsal joint at trot. Equine Vet J Suppl 2002;34:308–313. 26. Back W, Schamhardt HC, Savelberg HHCM. How the horse moves: significance of graphical representations of equine hind limb kinematics. Equine Vet J Suppl 1995;27:39–45. 27. van Weeren PR, van den Bogert AJ, Barneveld A. A quantitative analysis of skin displacement in the trotting horse. Equine Vet J Suppl 1990;9:101–109.

AJVR • Vol 79 • No. 2 • February 2018