Evaluation of discriminant analysis based on dorsoventral symmetry indices to quantify hindlimb lameness during over ground locomotion in the horse

Reasons for performing study: Advances in gait analysis techniques have led to assessment tools that can aid in detecting and quantifying lameness; here, bilateral tuberà coxae and pelvic movement during over ground locomotion are compared in order to investigate a practical method to assess hindlimb lameness in the horse. Objectives: To evaluate which parameters from anatomical landmarks on trunk and proximal hindlimbs are the best indicators of degree and side of hindlimb lameness. Methods: Fifteen horses (age 11–23 years, 6 nonlame and 9 unilaterally hindlimb lame horses 1/10 to 2/10 lame) were fitted with 4 inertial sensors: tuber sacrale, left and right tubera coxae and withers; 889 strides were collected from 6 trot trials per horse. Horses were assessed for lameness by a qualified equine orthopaedic surgeon from videos. Vertical displacement data for each sensor were used to calculate symmetry indices as well as published Fourier analysis based parameters. Linear discriminant analysis was used to determine the most discriminative parameters for 2 scenarios: grading of severity of lameness and identification of the affected limb. Results: Pelvic energy ratio gave the best indication for the degree of lameness. Directional symmetry index of the tubera coxae sensors yielded the highest discriminative power for identification of the lame limb. Conclusions and potential relevance: A good indication of the degree of hindlimb lameness can be obtained from vertical displacement data of the pelvic midline, collected from inertial sensors during over ground locomotion. The trunk mounted inertial sensor system allows for a time efficient collection of a representative database from horses with differing grade and site of lameness in a clinical setting. This is crucial for future work on a robust definition of the best parameters for lameness classification under practical conditions.     

The effect of induced hindlimb lameness on thoracolumbar kinematics during treadmill locomotion

REASON FOR PERFORMING STUDY: There are no detailed studies describing a relationship between hindlimb lameness and altered motion of the back. OBJECTIVES: To quantify the effect of induced subtle hindlimb lameness on thoracolumbar kinematics in the horse. METHODS: Kinematics of 6 riding horses were measured during walk and trot on a treadmill before and during application of pressure on the sole of the left hindlimb using a well-established sole pressure model. Reflective markers were located at anatomical landmarks on the limbs, back, head and neck for kinematic recordings. Ground reaction forces (GRF) in individual limbs were calculated from kinematics to detect changes in loading of the limbs. RESULTS: When pressure on the sole of the hindlimb was present, horses were judged as lame (grade 2 on the AAEP scale 1-5) by an experienced clinician. No significant unloading of this limb was found in the group of horses (unloading was observed in 4 animals, but was not detectable in the other 2), but statistically significant effects on back kinematics were detected. The overall flexion-extension (FE) range of motion (ROM) of the vertebral column was increased at walk, especially in the thoracic segments. Axial rotation (AR) ROM of the pelvis was also increased. At trot, the FE ROM was decreased only in the segment L3-L5-S3. During the stance phase of the lame limb, the segment T6-T10-T13 was more flexed and the neck was lowered at both gaits; the thoracolumbar segments were more extended at walk and trot. There were no significant changes in the stride length or protraction-retraction angles in any of the limbs. CONCLUSIONS: Subtle hindlimb lameness provoked slight but detectable changes in thoracolumbar kinematics. The subtle lameness induced in this study resulted in hyperextension and increased ROM of the thoracolumbar back, but also in decreased ROM of the lumbosacral segment and rotational motion changes of the pelvis. POTENTIAL RELEVANCE: Even subtle lameness can result in changes in back kinematics, which emphasises the intricate link between limb function and thoracolumbar motion. It may be surmised that, when chronically present, subtle lameness induces back dysfunction.     

Comparison of an inertial sensor system of lameness quantification with subjective lameness evaluation

Reasons for performing study: Subjective evaluation of mild lameness has been shown to have poor interobserver reliability. Traditional methods of objective lameness evaluation require specialised conditions and equipment. Wireless inertial sensor systems have been developed to allow for simple, rapid, objective lameness detection in horses trotted over ground. Objective: The purpose of this study was to compare the sensitivities of an inertial sensor system and subjective evaluation performed by experienced equine practitioners at detecting lameness in horses. We hypothesised that the inertial sensor system would identify lameness at a lower level of sole pressure than a consensus of 3 experienced equine veterinarians. Methods: Fifteen horses were fitted with special shoes that allowed for lameness induction via sole pressure. Horses were simultaneously evaluated by 3 equine veterinarians and a wireless inertial sensor system. Horses were subjected to multiple trials: 1) before inserting the screw; 2) after inserting the screw to just touch the sole; and 3) after tightening the screw in half turn increments. The number of screw turns required for lameness identification in the correct limb by the inertial sensors and by consensus of 3 equine veterinarians was compared using the Wilcoxon test. Results: The inertial sensor system selected the limb with the induced lameness after fewer screw turns than did the 3 veterinarians (P<0.0001). The inertial sensor system selected the correct limb before the 3 veterinarians in 35 trials (58.33%), the evaluators selected the correct limb before the inertial sensors in 5 trials (8.33%), and in 20 trials (33.33%) they selected the correct limb at the same time. Potential relevance: The inertial sensor system was able to identify lameness at a lower level of sole pressure than the consensus of 3 equine veterinarians. The inertial sensor system may be an effective aid to lameness localisation in clinical cases.     

Agreement between two inertial sensor gait analysis systems for lameness examinations in horses

Gait analysis is gaining popularity for quantification of lameness and 2 commonly used inertial sensor systems assess trunk movement symmetry: can these be used interchangeably in multi‐centre studies? We compared head and pelvic movement symmetry between 2 inertial sensor gait analysis systems in 13 horses equipped simultaneously with the 2 systems. The first system quantified dorsoventral movement in the local reference frame (System A) and the second system global vertical movement (System B). Widths of limits of agreement were calculated employing a well established regression method dealing with systematically changing differences over the range of measured values. Widths of limits of agreement between Systems A and B were narrower for pelvic movement than head movement. For head movement, they ranged from 6.4 to 6.9 mm for in‐hand trot and from 7.3 to 9.7 mm on the lunge and for pelvic movement from 2.5 to 4.4 mm in‐hand and from 3.6 to 5.3 mm on the lunge. Widths of limits of agreement between the 2 investigated inertial sensor gait analysis systems are of comparable magnitude (some equivalent, some marginally higher) to the currently proposed thresholds of 6 mm for head and 3 mm for pelvic movement used in lameness investigations. Differences in measurements with 2 different systems (A and B) obtained from the same horse falling within the reported values should not be seen as a sign of a change in lameness.     

Back kinematics of healthy trotting horses during treadmill versus over ground locomotion

Reasons for performing study: Treadmill locomotion is frequently used for training of sport horses, for diagnostic purposes and for research. Identification of the possible biomechanical differences and similarities between the back movement during treadmill (T) and over ground (O) locomotion is essential for the correct interpretation of research results. Objectives: To compare the kinematics of the thoracolumbar vertebral column in treadmill and over ground locomotion in healthy horses. Methods: Six sound Dutch Warmblood horses trotted on a T and O during 10 s at their own preferred velocity (mean ± s.d. 3.6 ± 0.3 m/s T and 3.6 ± 0.1 m/s O), which was the same in both conditions. Kinematics of the vertebral column was captured by infrared cameras using reflective skin markers attached over the spinous processes of selected vertebrae and other locations. Flexion‐extension and lateral bending range of motion (ROM), angular motion pattern (AMP) and intravertebral pattern symmetry (IVPS) of 5 vertebral angles (T6‐T10‐T13, T10‐T13‐T17, T13‐T17‐L1, T17‐L1‐L3 and L1‐L3‐l5) were calculated. Neck angle, linear and temporal stride parameters and protraction‐retraction angles of the limbs were also calculated. Results: The vertical ROM (flexion‐extension) was similar in both conditions, but the horizontal ROM (lateral bending) of the lumbar angles T17‐L1‐L3 and L1‐L3‐L5 was less during T locomotion (mean ± s.d. difference of 1.8 ± 0.6 and 1.7 ± 0.9°, respectively, P>0.05). During O locomotion, the symmetry pattern of the lumbar vertebral angles was diminished from 0.9 to 0.7 (1 = 100% symmetry) indicating increased irregularity of the movement (P>0.05). No differences were found in the basic linear and temporal stride parameters and neck angle. Potential relevance: Vertebral kinematics during treadmill locomotion is not identical to over ground locomotion, but the differences are minor. During treadmill locomotion lumbar motion is less, and caution should be therefore taken when interpreting lumbar kinematics.     

Interval between detection of lameness by locomotion scoring and treatment for lameness: A survival analysis

Estimates of point prevalence suggest that locomotion scoring identifies three times as many lame cows than when estimated by farmers. The aim of this study was to ascertain the impact of this under-recognition on the interval between identification of lameness (using locomotion score) and treatment. The study was undertaken on a 463-cow, spring-calving, pasture-fed herd in the lower North Island of New Zealand. All cows were locomotion scored (using a 1-5 scale) weekly during one milking season (July 2008 to May 2009). Survival analyses were then used to quantify the number of days between identification of a specific locomotion score and presentation, by farm staff, of a cow for lameness treatment. All cows which had a locomotion score of >3 were presented for lameness treatment subsequently, although >40% were treated more than 3weeks after being identified. Only 75% of events where cows had a locomotion score of 3 were followed by treatment with >65% of those treatments occurring >3weeks after the first score of 3. Improving the recognition of lameness by farm staff is thus likely to appreciably reduce the interval between reduced mobility and lameness treatment.     

The effect of induced forelimb lameness on thoracolumbar kinematics during treadmill locomotion

REASONS FOR PERFORMING STUDY: Lameness has often been suggested to result in altered movement of the back, but there are no detailed studies describing such a relationship in quantitative terms. OBJECTIVES: To quantify the effect of induced subtle forelimb lameness on thoracolumbar kinematics in the horse. METHODS: Kinematics of 6 riding horses was measured at walk and at trot on a treadmill before and after the induction of reversible forelimb lameness grade 2 (AAEP scale 1-5). Ground reaction forces (GRF) for individual limbs were calculated from kinematics. RESULTS: The horses significantly unloaded the painful limb by 11.5% at trot, while unloading at walk was not significant. The overall flexion-extension range of back motion decreased on average by 0.2 degrees at walk and increased by 3.3 degrees at trot (P<0.05). Changes in angular motion patterns of vertebral joints were noted only at trot, with an increase in flexion of 0.9 degrees at T10 (i.e. angle between T6, T10 and T13) during the stance phase of the sound diagonal and an increase in extension of the thoracolumbar area during stance of the lame diagonal (0.7degrees at T13, 0.8 degres at T17, 0.5 degres at L1, 0.4 degrees at L3 and 0.3 degrees at L5) (P<0.05). Lameness further caused a lateral bending of the cranial thoracic vertebral column towards the lame side (1.3 degrees at T10 and 0.9 degrees at T13) (P<0.05) during stance of the lame diagonal. CONCLUSIONS: Both range of motion and vertebral angular motion patterns are affected by subtle forelimb lameness. At walk, the effect is minimal, at trot the horses increased the vertebral range of motion and changed the pattern of thoracolumbar motion in the sagittal and horizontal planes, presumably in an attempt to move the centre of gravity away from the lame side and reduce the force on the affected limb. POTENTIAL RELEVANCE: Subtle forelimb lameness affects thoracolumbar kinematics. Future studies should aim at elucidating whether the altered movement patterns lead to back and/or neck dysfunction in the case of chronic lameness.     

Assessment of dairy cow locomotion in a commercial farm setting: The effects of walking speed on ground reaction forces and temporal and linear stride characteristics

Objectives of this study were to determine the effects of walking speed on ground reaction force (GRF) parameters and to explore inter- and intra-individual variability with unsupervised data collection in a commercial farm setting.We used eight high producing loose-housed Holstein Friesian cows consistently scored sound, with no veterinary treatment during the collection period. Cows walked freely (0.52-1.37 m/s) over a five force platform array, twice daily, following milking. GRF data were split into speed categories and temporal and kinetic gait parameters extracted. A general linear model was carried out to determine effects of speed.Variation in parameters between cows is inconsistent, while between speed categories (containing data from multiple cows) the parameters which vary are consistent. Stance and stride time were reduced with increasing speed but no change in peak vertical GRF or duty factor was found. This ability to track parameters within an individual over time aids detection of subtle changes associated with lameness.     

Kinematic analysis of the gait of 10 labrador retrievers during treadmill locomotion

The trotting gait of 10 sound, adult labrador retrievers was analysed using kinematic gait analysis on a purpose-built treadmill using video-based motion analysis software. The maximal angular displacement, minimal angular displacement, average angular displacement, and the maximal positive and negative angular velocities of the right elbow and right stifle were measured over five gait cycles at defined time points during each of five two-minute sessions. The dogs' trotting gait was not repeatable, either for individual dogs during the first session or between sessions, or between dogs at the same time points during a session.     

Evaluation of the sexual dimorphism in Mangalarga Marchador horses using discriminant analysis

This study used discriminant analysis to evaluate body measures and sexual dimorphism in Mangalarga Marchador horses. Discriminant analysis is a multivariate method that generates functions to classify animals using a prior criterion. In this study a prior criterion is male or female. The study analyzed 25 linear and 11 angular measures obtained from 25 stallions and 56 mares registered with the Mangalarga Marchador Brazilian Breeders Association. Knee girth, cannon girth, chest width, chest girth, back-loin length, hip width, distance from elbow to knee, hind limb hoof length, head width, shoulder length, and body length were found to be linear measures that provided evidence of sexual dimorphism. Angular measures with evidence of sexual dimorphism were shoulder–humerus, metacarpal–phalanx, coxae–femur, and femur–tibial angles. Stallions were larger than mares for almost all linear measures, except for back-loin length, hip width, and distance from elbow to knee. Shoulder–humerus and coxal–femur angles were larger in mares, while metacarpal–phalanx and femur–tibial angles were larger in stallions. Using linear measures, two mares were classified as stallions and three stallions were classified as mares. For angular measures, two other mares were classified as stallions and three other stallions were classified as mares. The discriminant functions can be useful for the selection of horses and to avoid registration of animals not meeting the phenotypic standards of the breed association.     

Phytomass estimation using eMODIS NDVI and ground data in arid rangelands of Morocco

The assessment of rangeland productivity in semi-extensively grazed arid rangelands is a prerequisite for livestock management in relation to sustainable use of pastoral resources. The objective of this study was to assess rangeland productivity based on normalised difference vegetation index (NDVI) images. Data on phytomass were measured on 61 field samples in arid rangelands of Morocco, covering various rangeland categories during autumn (November) and spring (April), i.e. when phytomass is at low and high levels, respectively, for two consecutive years (2008 and 2009). Dekadal EROS Moderate Resolution Imaging Spectroradiometer (eMODIS) NDVI data were linearly regressed to field measurements for these four periods. Results show that phytomass values were correlated with NDVI during spring, with R² and RMSE values of 0.82 and 0.3 t ha^?1, respectively. This study indicates there is a high potential for operational use of remotely sensed data to estimate rangeland phytomass of semi-extensively grazed rangelands.     

Disentangling data dependency using cross-validation strategies to evaluate prediction quality of cattle grazing activities using machine learning algorithms and wearable sensor data

Wearable sensors have been explored as an alternative for real-time monitoring of cattle feeding behavior in grazing systems. To evaluate the performance of predictive models such as machine learning (ML) techniques, data cross-validation (CV) approaches are often employed. However, due to data dependencies and confounding effects, poorly performed validation strategies may significantly inflate the prediction quality. In this context, our objective was to evaluate the effect of different CV strategies on the prediction of grazing activities in cattle using wearable sensor (accelerometer) data and ML algorithms. Six Nellore bulls (average live weight of 345 ± 21 kg) had their behavior visually classified as grazing or not-grazing for a period of 15 d. Elastic Net Generalized Linear Model (GLM), Random Forest (RF), and Artificial Neural Network (ANN) were employed to predict grazing activity (grazing or not-grazing) using 3-axis accelerometer data. For each analytical method, three CV strategies were evaluated: holdout, leave-one-animal-out (LOAO), and leave-one-day-out (LODO). Algorithms were trained using similar dataset sizes (holdout: n = 57,862; LOAO: n = 56,786; LODO: n = 56,672). Overall, GLM delivered the worst prediction accuracy (53%) compared with the ML techniques (65% for both RF and ANN), and ANN performed slightly better than RF for LOAO (73%) and LODO (64%) across CV strategies. The holdout yielded the highest nominal accuracy values for all three ML approaches (GLM: 59%, RF: 76%, and ANN: 74%), followed by LODO (GLM: 49%, RF: 61%, and ANN: 63%) and LOAO (GLM: 52%, RF: 57%, and ANN: 57%). With a larger dataset (i.e., more animals and grazing management scenarios), it is expected that accuracy could be increased. Most importantly, the greater prediction accuracy observed for holdout CV may simply indicate a lack of data independence and the presence of carry-over effects from animals and grazing management. Our results suggest that generalizing predictive models to unknown (not used for training) animals or grazing management may incur poor prediction quality. The results highlight the need for using management knowledge to define the validation strategy that is closer to the real-life situation, i.e., the intended application of the predictive model.     

A descriptive analysis of the foot lesions identified during veterinary treatment for lameness on dairy farms in north Taranaki

AIM: To analyse the records made during routine veterinary treatment of lame cows, to identify the types of lesions causing lameness of dairy cows in north Taranaki, their relative prevalence and distribution, and some of the cow-associated factors which affect these parameters. METHODS: Data concerning cow age and breed, lesion type, lesion site, and whether the examination was a revisit were collected from records of veterinary visits between December 1995 and July 2007. Data were analysed on a single- or two-factor basis, using chi(2) and, where suitable, calculation of OR. RESULTS: The total number of individual cow treatments recorded was 4,488. The most commonly recorded lesion was white-line disease (WLD) (42% of diagnoses at first examination), and the second was sole injury (29%). Heifers accounted for 19% of cases, similar to the expected replacement rate. Jerseys accounted for 18% of cases, lower than the proportion of purebred Jerseys in the region (27%). There was a significant effect of age and breed on the prevalence and distribution of lesions. There were also significant differences between lesion types in their location, which were also influenced by breed and age. CONCLUSIONS: This study provided a comprehensive overview of the lesions causing lameness in dairy cattle in north Taranaki, and identified some of the factors which influence the development of lesions. More research is needed to better establish the aetiology and pathogenesis of lameness in cattle under New Zealand conditions. CLINICAL RELEVANCE: This study provides baseline data with which veterinarians can compare their results to identify significant patterns on specific farms, which will suggest specific causal factors for lameness on those farms and thus better guide prevention programmes.