- Email: [email protected]
Thermographic Evaluation of Racehorse Performance
Journal of Equine Veterinary Science 34 (2014) 1076–1083
Contents lists available at ScienceDirect
Journal of Equine Veterinary Science journal homepage: www.j-evs.com
Original Research
Thermographic Evaluation of Racehorse Performance Maria Soroko PhD a, *, Krzysztof Dudek PhD b, Kevin Howell PhD c, Ewa Jodkowska a, Radomir Henklewski PhD d a
Department of Horse Breeding and Equestrian Studies, Wroclaw University of Environmental and Life Sciences, Wroclaw, Poland Institute of Machines Design and Operation, Technical University of Wroclaw, Wroclaw, Poland c Microvascular Diagnostics, Institute of Immunity and Transplantation, Royal Free Hospital, London, UK d Department of Surgery, Wroclaw University of Environmental and Life Sciences, Wroclaw, Poland b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 8 April 2014 Received in revised form 28 May 2014 Accepted 13 June 2014 Available online 25 June 2014
The present study was aimed at identifying the key thermographic diagnostic areas essential for monitoring the effect of training on racehorses. The study involved monitoring 15 racehorses in 13 imaging sessions over a period of 10 months. Temperature measurements were made at a total of 46 regions of interest (ROIs) at the distal parts of the limbs and the back. To account for the influence of ambient temperature on each ROI measurement, values were adjusted to a constant ambient temperature of 12 C, estimated using regression analysis. The horses in the study were divided into two groups based on the value of success rate in racing competition. During the research period, none of the horses were identified as injured by routine veterinary investigation. Successful horses had significantly warmer adjusted ROI temperatures than their less successful counterparts at both carpal joints, the third metacarpal bones, the left fetlock joint, the left front short pastern bone, the left tarsus joint, and the caudal part of the thoracic vertebrae. The study tested a protocol for recording body surface temperature in racehorses, which was shown to increase reliability by adjusting for variations in ambient temperature. When analyzed on the basis of sporting performance, the protocol identified 14 ROIs that were associated with superior performance, most of which were at the limbs on the left side. Ó 2014 Elsevier Inc. All rights reserved.
Keywords: Thermography Racehorse Sport performance Training
1. Introduction The health and training performance level of horses can be evaluated by behavioral, hematological, and biochemical changes in the body during physiological effort. Previous studies have investigated basic physiological indicators such as blood chemistry and pulse rate [1–3], but they have not provided evidence of the changes to the equine musculoskeletal system that occur in response to regular training. Jodkowska [4] identified the possibility of monitoring the impact of training through the measurement of body * Corresponding author at: Maria Soroko, PhD, Department of Horse Breeding and Equestrian Studies, Wroclaw University of Environmental and Life Sciences, Kozuchowska 5A, 51-161 Wroclaw, Poland. E-mail address: [email protected] (M. Soroko). 0737-0806/$ – see front matter Ó 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jevs.2014.06.009
surface temperature, which is influenced by the vascularity and metabolic activity within and below the skin surface. Thermography is a noninvasive measurement method enabling abnormal patterns of skin surface temperature to be detected by means of infrared imaging. Potential veterinary applications of thermography have been described [5–8]. These reports mostly address the detection of injuries caused by training overloads, especially at the back and the distal parts of the limbs of the horse [9–11]. Normal thermographic patterns in the horse have also been described, enabling improved diagnostic interpretation of equine thermography [5,12,13]. To date, however, no study has located the optimal body regions for reliable thermographic evaluation of musculoskeletal impact on horses. Therefore, the present study was aimed at identifying the key thermographic diagnostic areas essential for monitoring the effect of training on racehorses.
M. Soroko et al. / Journal of Equine Veterinary Science 34 (2014) 1076–1083
1077
2. Materials and Methods 2.1. Study Population and Data Collection The study was approved by the Local Ethical Committee for Experiments on Animals of Wroclaw University of Environmental and Life Sciences. Thermography measurements were obtained from 15 clinically healthy racehorses of two breeds (12 Polish Half Breed and three Arabians), all aged 3 years. All horses were trained for flat racing at Partynice Race Course (Poland) during the 2012 season. Performance data such as orthopedic injuries, training intensity, and success rate (SR, prize money per race) for 2012 were collected for each horse. The horses were housed in individual stalls with common management and training regimes. All horses were trained and raced in a clockwise direction. The protocol for thermography was as previously described by Van Hoogmoed et al [14] and Soroko et al [15]. Horses were examined at rest before daily exercise. Dirt and mud present in the imaging field of view was brushed away 15 minutes before examination and approximately 10 minutes were allowed to pass before scanning to ensure the transient heat generated by brushing had subsided before obtaining baseline measurements. The brushing was performed symmetrically. To minimize the effect of environmental factors, thermography was always performed at the same place within an enclosed stable. The distance of the animal from the camera was fixed for all imaging at 1 m, and the emissivity (3) was set to 1 for all readings. At each imaging session, a series of six thermographic images were taken of the dorsal, lateral, and medial aspects of the distal part of the forelimbs (Figs. 1–3); the dorsal aspect of the distal part of the hind limbs (Figs. 4 and 5); and the dorsal aspect of the back (Fig. 6) using a VarioCam HR infrared camera (uncooled microbolometer focal plane array; resolution, 640 480 pixels; spectral range,
7.5–14 mm; InfraTec, Dresden, Germany). A total of 13 imaging sessions were conducted over a period of 10 months. At each session, the ambient temperature in the stable (Tamb) was measured by a TES 1314 thermometer (TES, Taipei, Taiwan). Regions of interest (ROIs) were determined for each thermographic image. On each image of the forelimbs 10 ROIs were defined (Figs. 1–3), allowing temperature measurements to be made bilaterally at the dorsal, lateral, and medial aspects of the carpal joint, third metacarpal bone, fetlock joint and short pastern bone; the dorsal aspect of the hoof; and the lateral and medial aspects of the heel. On each image of the hind limbs 5 ROIs were defined (Figs. 4 and 5) allowing bilateral measurements at the
Fig. 1. Thermogram of dorsal aspect of distal part of forelimbs. Measured regions of interest: DF1dright carpal joint, dorsal aspect; DF2dleft carpal joint, dorsal aspect; DF3dright third metacarpal bone, dorsal aspect; DF4dleft third metacarpal bone, dorsal aspect; DF5dright fetlock joint, dorsal aspect; DF6dleft fetlock joint, dorsal aspect; DF7dright short pastern bone, dorsal aspect; DF8dleft short pastern bone, dorsal aspect; DF9dright hoof, dorsal aspect; and DF10dleft hoof, dorsal aspect.
Fig. 3. Thermogram of left lateral and right medial aspects of distal part of the forelimbs. Measured regions of interest: MF1dright carpal joint, medial aspect; LF2dleft carpal joint, lateral aspect; MF3dright third metacarpal bone, medial aspect; LF4dleft third metacarpal bone, lateral aspect; MF5dright fetlock joint, medial aspect; LF6dleft fetlock joint, lateral aspect; MF7dright short pastern bone, medial aspect; LF8dleft short pastern bone, lateral aspect; MF9dright heel, medial aspect; and LF10dleft heel, lateral aspect.
Fig. 2. Thermogram of right lateral and left medial aspects of distal part of the forelimbs. Measured regions of interest: LF1dright carpal joint, lateral aspect; MF2dleft carpal joint, medial aspect; LF3dright third metacarpal bone, lateral aspect; MF4dleft third metacarpal bone, medial aspect; LF5dright fetlock joint, lateral aspect; MF6dleft fetlock joint, medial aspect; LF7dright short pastern bone, lateral aspect; MF8dleft short pastern bone, medial aspect; LF9dright heel, lateral aspect; and MF10dleft heel, medial aspect.
1078
M. Soroko et al. / Journal of Equine Veterinary Science 34 (2014) 1076–1083
Fig. 4. Thermogram of dorsal aspect of distal part of right hind limb. Measured regions of interest: DH1dright tarsus joint, dorsal aspect; DH3dright third metatarsal bone; dorsal aspect; DH5dright hind fetlock joint dorsal aspect; DH7dright hind short pastern bone, dorsal aspect; and DH9dright hind hoof, dorsal aspect.
dorsal aspect of the tarsus joint, the third metatarsal bone, fetlock joint, short pastern bone, and hoof. On the image of the spine 6 ROIs were defined (Fig. 6) to obtain measurements from the dorsal aspect of the cranial and caudal parts of the thoracic vertebrae; the left and right sides of the thoracic vertebrae; the lumbar vertebrae; and the sacroiliac joints. The temperature was calculated using IRBIS 3 Professional software (InfraTec, Dresden, Germany) using the mean pixel value in the circular ROI for joints, hoofs, and heels; the mean pixel value along the linear ROI for long bones; and the mean value within the polygon for ROIs at the back. The average temperature was obtained from 8,786 ROIs. Not all ROIs were always available for analysis
Fig. 6. Thermogram of dorsal aspect of back. Measured regions of interest: B1dcranial part of thoracic vertebrae, dorsal aspect; B2dcaudal part of thoracic vertebrae, dorsal aspect; BL1dleft side of thoracic vertebrae, dorsal aspect; BR1dright side of thoracic vertebrae, dorsal aspect; B3dlumbar vertebrae, dorsal aspect; and B4dsacroiliac joints.
due to premature withdrawal of the horse from the racing season or inaccessibility of the body region for imaging. For estimation of repeatability of the analysis, ROIs were placed twice by a single operator on the thermographic images collected in the same session in 10 randomly chosen horses. Every 3 months, standard veterinary methods were used for palpation examinations of the distal parts of the forelimbs, hind limbs and back. Additional ultrasonographic and radiographic examinations were performed for the forelimbs. These were conducted by a veterinarian to diagnose any pathologic conditions. For ultrasonographic examinations, an Echo Blaster 128 scanner (Telemed, Vilnius, Lithuania) was used, and for radiographic examinations, Gierth HF90 (Gierth, Riesa, Germany) and Conaxx 35 (PROTEC, Oberstenfeld, Germany) systems were used. 2.2. Statistical Analysis All statistical analysis was performed using STATISTICA v. 10 (StatSoft, Tulsa). 2.2.1. ROI Temperature Adjustment At each body site, the correlation of ROI temperature (TROI) with ambient temperature (Tamb) was analyzed. To account for the influence of ambient temperature on each ROI temperature measurement, TROI values were adjusted adj to a value TROI for a constant ambient temperature of 12 C, estimated using the equation:
adj 2 TROI ¼ TROI þ b1 ,ð12 Tamb Þ þ b2 , 122 Tamb Model coefficients, b1 and b2, were estimated for each ROI separately using regression analysis. An example of adjustment for two readings taken from the ROI DF1 is given in Fig. 7. Fig. 5. Thermogram of dorsal aspect of distal part of left hind limb. Measured regions of interest: DH2dleft tarsus joint, dorsal aspect; DH4dleft third metatarsal bone, dorsal aspect; DH6dleft hind fetlock joint, dorsal aspect; DH8dleft hind short pastern bone, dorsal aspect; and DH10dleft hind hoof, dorsal aspect.
2.2.2. Coefficient of Variation in Temperature Within Each ROI The range of pixel temperature values within an ROI depends on its location on the body of a horse (related to the musculature, vasculature, fat thickness, and so forth)
M. Soroko et al. / Journal of Equine Veterinary Science 34 (2014) 1076–1083
1079
(SR): group 1: SR V400/race, 12 horses (including eight mares) and 152 pooled measurement results and group 2: SR > V400/race, three horses (including two mares) and 39 pooled measurement results. The difference in mean adjusted temperature (DT) between group 2 horses (SR > V400) and group 1 horses (SR V400) was calculated for each ROI. Statistical significance was tested using Student t test, with results considered significant if P < .05. An example box-andwhisker plot for group 1 and group 2 horses is shown for ROI DF1 in Fig. 8.
Fig. 7. Example of temperature adjustment principle to ambient temperature Tamb ¼ 12 C for two readings at the region of interest DF1dright carpal joint, dorsal aspect. Tamb, ambient temperature; TDF1, unadjusted temperature of ROI DF1; TadjDF1, temperature of ROI DF1 adjusted to an ambient temperature of 12 C.
and also on the size of the surface area. A uniform temperature across the ROI will assist in improved repeatability of results and implies that the ROI size and position is physiologically relevant. The coefficient of variation (CV) was adopted as the benchmark of temperature uniformity within each ROI:
CV ¼
SD ,100; M
where SD is the standard deviation of the pixel temperature values and M is the mean temperature value within the ROI, both extracted from the IRBIS 3 Professional software output. The mean values of the CV (across all horses and all measurement sessions) are presented in Table 1. 2.2.3. Repeatability of the Results Expressed by Intraclass Correlation Coefficient rICC Measurement repeatability in each of the 46 ROI was based on intraclass correlation coefficients, rICC. The following formula was used [16]:
rICC ¼
MSBS MSres MSBS þ ðk 1Þ,MSres þ nk ðMSBC MSres Þ
where MSBCdmean squares between couples; MSBSd mean of squares between subjects; MSresdmean squared residual n–number of horses (n ¼ 10); kdnumber of measurements (k ¼ 2). The value of rICC z 1 indicates a strong absolute temperature consistency in the assessment of ROI by the evaluator. This is reflected in a large variance between the objects (a significant difference between average n measurements) and low variance among repeated measurements by the evaluator (an insignificant difference in average results among k repeated measurements). 2.2.4. Difference in Mean ROI Temperature Among Horses With Different Success Rates The average prize won by horses competing in the 2012 season was V400/race. The horses in the study were divided into two groups based on this value of success rate
2.2.5. Area Under the Receiver Operating Characteristic Curve Receiver operating characteristic (ROC) curve analysis [17] makes it possible to identify whether the test parameter (adjusted temperature) of the ROI allows the correct classification of the analyzed horse to one of the two groups (group 1, SR V400 or group 2, SR > V400) in the most reliable way. Receiver operating characteristic analysis identifies a “cutoff” temperature which optimizes the sensitivity and specificity of the test for classifying the horse into the correct group. An example for ROI LF2 is given in Fig. 9, wherein (100dspecificity) is plotted against sensitivity. A cutoff at the top-left corner of the plot would be able to group the analyzed horse with perfect sensitivity and specificity. In this case, the area under the curve (AUC) would be 1. Conversely, a cutoff lying along the dotted line of identity would have no discriminatory power for grouping the analyzed horse, and in this case, the AUC would be 0.5. The optimum cutoff for region LF2 is marked along the ROC curve with a dot, and the AUC is 0.782. 3. Results During the research period, none of the horses were identified as injured by the ultrasonographic, radiographic, or palpation examinations. The ambient temperature in the stable Tamb ranged between 2.8 C and 22.7 C (Fig. 7). Temperature data derived from the 46 ROIs are presented in Table 1. The table shows mean and standard deviation ROI temperature values ðT ROI SDÞ (all horses, all measurement sessions); minimum temperature recorded for the ROI (tmin) and the maximum value (tmax); mean CV of pixel temperature values within each ROI; intraclass correlation coefficients (rICC) for the repeated readings from 10 horses; correlation coefficient (R) of ROI temperature with Tamb and the values of the regression coefficients b1 and b2; mean adjusted ROI temperatures to a constant ambient temperadj ature of 12 C (all horses, all measurement sessions) ðT ROI Þ. The mean CV for pixel values within each ROI was generally low, with the highest mean CV being 6.4% at ROI LF6 (left fetlock joint, lateral aspect). Intraclass correlation coefficients (rICC) for the repeated readings from 10 horses were generally high: 38/46 ROIs had an rICC greater than 0.8. The lowest repeatability was observed at ROIs MF1 (right carpal joint, medial aspect), rICC ¼ 0.614, and MF2 (left carpal joint, medial aspect), rICC ¼ 0.626. Adjustment of the ROI temperatures using the regression analysis limited the influence of ambient temperature on each mean ROI reading. Standard deviations on the adjusted mean temperature readings were greatly reduced
1080
M. Soroko et al. / Journal of Equine Veterinary Science 34 (2014) 1076–1083
Table 1 Basic temperature statistics of analyzed regions of interest (ROIs). Region (ROI)
T ROI SD
DF1dright carpal joint, dorsal aspect DF2dleft carpal joint, dorsal aspect DF3dright third metacarpal bone, dorsal aspect DF4dleft third metacarpal bone, dorsal aspect DF5dright fetlock joint, dorsal aspect DF6dleft fetlock joint, dorsal aspect DF7dright short pastern bone, dorsal aspect DF8dleft short pastern bone, dorsal aspect DF9dright hoof, dorsal aspect DF10dleft hoof, dorsal aspect LF1dright carpal joint, lateral aspect MF1dright carpal joint, medial aspect LF2dleft carpal joint, lateral aspect MF2dleft carpal joint, medial aspect LF3dright third metacarpal bone, lateral aspect MF3dright third metacarpal bone, medial aspect LF4dleft third metacarpal bone, lateral aspect MF4dleft third metacarpal bone, medial aspect LF5dright fetlock joint, lateral aspect MF5dright fetlock joint, medial aspect LF6dleft fetlock joint, lateral aspect MF6dleft fetlock joint, medial aspect LF7dright short pastern bone, lateral aspect MF7dright short pastern bone, medial aspect LF8dleft short pastern bone, lateral aspect MF8dleft short pastern bone, medial aspect LF9dright heel, lateral aspect MF9dright heel, medial aspect LF10dleft heel, lateral aspect MF10dleft heel, medial aspect DH1dright tarsus joint, dorsal aspect DH2dleft tarsus joint, dorsal aspect DH3dright third metatarsal bone, dorsal aspect DH4dleft third metatarsal bone, dorsal aspect DH5dright hind fetlock joint, dorsal aspect DH6dleft hind fetlock joint, dorsal aspect DH7dright hind short pastern bone, dorsal aspect DH8dleft hind short pastern bone, dorsal aspect DH9dright hind hoof, dorsal aspect DH10dleft hind right hoof, dorsal aspect B1dcranial part of thoracic vertebrae, dorsal aspect B2dcaudal part of thoracic vertebrae, dorsal aspect BL1dleft side of thoracic vertebrae, dorsal aspect BR1dright side of thoracic vertebrae, dorsal aspect B3dlumbar vertebrae, dorsal aspect B4dsacroiliac joints, dorsal aspect
25.5 25.3 24.0 24.2 24.4 24.6 24.0 24.1 27.7 27.6 25.4 25.8 26.4 26.3 24.2 24.3 24.4 24.3 24.4 24.4 24.9 24.6 25.8 25.7 26.1 25.8 28.7 28.4 28.5 28.3 28.3 28.2 25.6 25.5 24.9 24.8 21.2 21.1 25.7 25.8 29.4 29.9 28.7 28.5 28.9 28.5
5.1 5.1 5.5 5.6 5.3 5.4 5.2 5.2 4.9 4.9 5.5 5.1 5.0 4.9 5.7 5.7 5.7 5.7 5.6 5.5 5.9 5.9 5.3 5.1 5.4 5.4 4.7 4.8 4.8 4.7 4.5 4.6 5.3 5.3 5.2 5.2 4.9 4.8 5.2 4.8 4.5 4.3 5.3 5.3 4.7 5.0
adj
tmin
tmax
Mean CV (%)
rICC
R
b1
b2
T ROI SD
9.9 10.5 7.6 7.8 8.1 7.7 7.9 7.2 9.1 9.0 11.7 12.1 12.6 12.1 8.4 9.1 8.4 8.3 7.6 8.1 9.0 8.0 9.6 9.2 8.8 8.2 10.6 10.8 9.1 9.4 12.5 14.5 8.6 10.2 7.2 9.6 6.7 7.9 10.4 11.4 18.4 19.1 15.2 14.3 16.0 16.4
32.5 32.6 32.6 32.8 32.5 32.4 31.4 31.4 34.0 33.9 33.5 34.1 33.5 34.2 32.4 32.4 32.6 32.6 32.5 31.7 32.4 32.5 32.9 33.0 33.5 33.2 35.0 34.7 34.8 34.4 34.2 33.7 32.5 32.4 31.8 31.8 29.4 29.9 32.5 32.5 35.1 35.4 35.1 34.8 35.0 34.9
2.8 2.8 2.3 2.4 2.9 3.2 3.3 3.2 2.8 3.1 4.3 5.6 3.6 5.8 4.8 5.5 4.6 4.9 6.3 5.9 6.4 5.8 4.4 5.3 4.1 5.7 5.2 5.4 5.0 5.0 3.0 3.2 5.5 2.9 3.1 5.1 5.5 5.9 4.3 4.2 2.3 2.1 3.0 2.9 1.6 1.6
0.920 0.937 0.981 0.960 0.951 0.943 0.940 0.913 0.942 0.922 0.810 0.614 0.682 0.626 0.892 0.861 0.949 0.959 0.896 0.878 0.882 0.868 0.965 0.923 0.934 0.949 0.878 0.809 0.815 0.881 0.920 0.923 0.860 0.898 0.955 0.954 0.777 0.789 0.722 0.825 0.849 0.803 0.773 0.770 0.970 0.985
0.805 0.789 0.788 0.772 0.788 0.741 0.805 0.746 0.630 0.566 0.845 0.847 0.819 0.828 0.847 0.844 0.839 0.828 0.856 0.852 0.801 0.805 0.812 0.798 0.767 0.784 0.704 0.694 0.613 0.624 0.809 0.790 0.774 0.749 0.759 0.746 0.593 0.584 0.351 0.394 0.882 0.878 0.896 0.889 0.897 0.900
0.84 0.82 0.93 0.92 0.89 0.86 0.77 0.73 0.52 0.38 0.87 0.94 0.87 0.45 0.46 1.02 1.04 0.43 0.44 0.96 1.02 0.47 0.44 0.88 0.82 0.43 0.40 0.78 0.33 0.62 1.08 1.10 1.01 1.03 0.90 0.97 0.68 0.64 0.40 0.35 0.99 1.00 1.07 1.10 1.27 1.27
0.001 0 0.002 0.002 0.002 0.001 0.002 0.003 0.006 0.010 0.002 0.006 0.003 0.010 0.014 0.004 0.005 0.015 0.015 0.003 0.004 0.013 0.013 0.002 0.001 0.012 0.009 0.003 0.011 0.001 0.012 0.013 0.006 0.006 0.004 0.006 0.002 0.004 0.006 0.010 0.009 0.011 0.010 0.010 0.015 0.015
23.0 23.0 21.7 21.8 22.1 22.3 21.7 21.8 25.6 25.5 24.4 24.6 24.8 25.0 21.8 22.0 22.0 21.8 21.9 22.0 22.3 21.9 23.2 23.6 23.3 23.0 26.8 26.5 28.0 28.2 26.5 26.4 22.5 22.7 22.7 22.2 22.1 22.1 26.4 26.4 27.9 28.2 26.7 26.6 27.0 26.8
2.3 2.3 2.6 2.7 2.4 2.7 2.3 2.6 3.0 3.3 2.1 2.0 2.1 2.2 2.4 2.2 2.3 2.5 2.3 2.1 2.6 2.8 2.4 2.3 2.6 2.6 3.1 2.7 3.0 3.0 2.0 2.1 2.5 2.5 2.7 3.2 3.1 3.0 4.2 3.7 1.5 1.4 1.6 1.7 1.5 1.5
b1, b2, binomial regression coefficients of ROI with Tamb; Mean CV, mean coefficient of variation of pixel temperature values for each ROI; R, correlation coefficient of ROI temperature with ambient temperature (Tamb); rICC, intraclass correlation coefficient for repeatability of readings (two repeat readings, 10 horses); tmax, maximum temperature value ( C); tmin, minimum temperature value ( C); T ROI SD, mean and standard deviation of temperature in ROI ( C); adj T ROI SD, mean and standard deviation of adjusted temperature in ROI ( C).
compared with SDs for the unadjusted data. Adjusted ROI temperatures showed a generally high degree of symmetry between the left and right sides. Asymmetry for all sites was less than 1 C, with the exception of the lateral and medial aspects of the front heels. The ROC and Student t test analysis of SR is detailed in Table 2. Successful (group 2) horses had warmer adjusted ROI temperatures than their less successful counterparts (group 1) at all body sites. Differences in adjusted temperature between the groups were significantly higher (P < .05) in 14 of 46 sites (bold values in the table), including all aspects of both carpal joints, some views of the third metacarpal bones, the lateral and medial aspects of the left fetlock joint, the lateral aspect of the left front short
pastern bone, the dorsal aspect of the left tarsus joint, and the caudal part of the thoracic vertebrae. The highest AUC values were registered in the adj following regions: TLF2 dleft carpus joint, medial aspect adj dleft fetlock joint, lateral aspect (AUC ¼ 0.782); TLF6 adj dleft third metacarpal bone, medial (AUC ¼ 0.763); and TLF4 aspect (AUC ¼ 0.733).
4. Discussion Thermography has been most commonly used to evaluate horses with limb lameness. Clinical diseases that have been recognized and characterized by thermography
M. Soroko et al. / Journal of Equine Veterinary Science 34 (2014) 1076–1083
Fig. 8. Example of comparison of adjusted temperatures of horses from groups 1 and 2 for region of interest DF1dright carpal joint, dorsal aspect. SE, standard error; SR, success rate; TadjDF1, temperature of ROI DF1 adjusted to an ambient temperature of 12 C.
include tendonitis, laminitis, and sole abscesses. It has also been successfully used for back abnormality diagnosis associated with neuromuscular disease of the thoracolumbar region and inflammation of spinous processes [8,18–20]. Most racehorse injures are associated with the physical demands on the musculoskeletal system in response to training [21], and they can also be linked to the type of training and the skills of the rider [22,23]. The detection and monitoring of physiological responses to training overloads are particularly important for racehorses put under extreme physical demands. In the study presented by Jeffcott et al [24], of 163 Thoroughbred racehorses, 53% suffered from lameness, which was the main cause of elimination of the horses from sport performance. A more recent study involving Thoroughbred racehorses recorded an 81% incidence of lameness [25]. Thermography has been used to identify areas of inflammation that could decrease racehorse performance [26]. In another study, it was found that the temperature of the back and the distal parts of the forelimbs in racehorses at rest was increased by long-term training [27]. In the present study, 10 months of regular thermographic examination of racehorses allowed the identification
1081
of the most important body regions to be recommended for monitoring the impact of training. Identifying the key thermographic diagnostic areas should facilitate the detection of pathologic conditions during the training cycle. This is particularly important in racehorses, where immediate diagnosis might help to maintain their health and condition, positively influencing their further career in sport. The determination of the ROIs we used was based on thermographic measurement standards in veterinary medicine. Repeatability of the thermographic measurements was generally good, with the notable exception of the medial aspect of both carpal joints. This poor repeatability may arise because these ROIs must be imaged at an angle around the contralateral limb, which might limit reproducibility of the imaging technique. We introduced an adjustment to take into consideration the effect of ambient temperature on body surface temperature; this reduced variability in the measurements and allowed ROIs that conveyed the best information about racehorse performance to be more reliably identified. We observed no correlation (P > .05) between the ambient temperature at which the horses were studied and SR. Hence the process of adjusting ROIs for ambient temperature was unlikely to systematically influence the study’s ability to classify horse performance by analyzing ROI temperatures. Our study has a number of important limitations. First, we were not able to collect data at consistent time points after each race, and this may reduce the ability of the study to classify the horses’ performance on the basis of ROI temperature. Second, we used the prize money earned by each horse as a measure of performance. Prize earnings are of course influenced not just by horse performance but by many additional factors such as the jockey and the standard of competition. Third, the number of horses in our study was small, with a bias toward poorer performing horses. A larger study with an even distribution of performance would increase the validity of the thermographic findings. None of the horses were injured during the measurement period, but nonetheless, the best performing racehorses (group 2) were warmer than their poorer performing peers (group 1) at all body sites studied. This temperature difference may reflect an increased physiological training impact on the best performing horses. Further work is required to establish if our protocol can be
adj Fig. 9. Receiver operating characteristic curve for LF2dleft carpal joint, lateral aspect; cutoff point for TLF2 > 24.4 C; and histogram for groups 2 and 1. AUC, area under the curve; ROI, region of interest; Sens, sensitivity; Spec, specificity; Tadj, temperature of ROI adjusted to an ambient temperature of 12 C.
1082
M. Soroko et al. / Journal of Equine Veterinary Science 34 (2014) 1076–1083
Table 2 Receiver operating characteristic (ROC) analysis and Student t test results for discrimination of horses in group 2 (SR > V400) from those in group 1 (SR V400). Region (ROI)
DF1dright carpal joint, dorsal aspect DF2dleft carpal joint, dorsal aspect DF3dright third metacarpal bone, dorsal aspect DF4dleft third metacarpal bone, dorsal aspect DF5dright fetlock joint, dorsal aspect DF6dleft fetlock joint, dorsal aspect DF7dright short pastern bone, dorsal aspect DF8dleft short pastern bone, dorsal aspect DF9dright hoof, dorsal aspect DF10dleft hoof, dorsal aspect LF1dright carpal joint, lateral aspect MF1dright carpal joint, medial aspect LF2dleft carpal joint, lateral aspect MF2dleft carpal joint, medial aspect LF3dright third metacarpal bone, lateral aspect MF3dright third metacarpal bone, medial aspect, LF4dleft third metacarpal bone, lateral aspect MF4dleft third metacarpal bone, medial aspect LF5dright fetlock joint, lateral aspect MF5dright fetlock joint, medial aspect LF6dleft fetlock joint, lateral aspect MF6dleft fetlock joint, medial aspect LF7dright short pastern bone, lateral aspect MF7dright short pastern bone, medial aspect LF8dleft short pastern bone, lateral aspect MF8dleft short pastern bone, medial aspect LF9dright heel, lateral aspect MF9dright heel, medial aspect LF10dleft heel, lateral aspect MF10dleft heel, medial aspect DH1dright tarsus joint, dorsal aspect DH2dleft tarsus joint, dorsal aspect DH3dright third metatarsal bone, dorsal aspect DH4dleft third metatarsal bone, dorsal aspect DH5dright hind fetlock joint, dorsal aspect DH6dleft hind fetlock joint, dorsal aspect DH7dright hind short pastern bone, dorsal aspect DH8dleft hind short pastern bone, dorsal aspect DH9dright hind hoof, dorsal aspect DH10dleft hind right hoof, dorsal aspect B1dcranial part of thoracic vertebrae, dorsal aspect B2dcaudal part of thoracic vertebrae, dorsal aspect BL1dleft side of thoracic vertebrae, dorsal aspect BR1dright side of thoracic vertebrae, dorsal aspect B3dlumbar vertebrae, dorsal aspect B4dsacroiliac joints, dorsal aspect
ROC analysis
Group 2 Versus Group 1
Cutoff ( C)
Sensitivity (%)
Specificity (%)
AUC
DT ( C)
P
>23.4 >23.4 >23.2 >22.4 >23.2 >22.9 >23.6 >22.5 >24.9 >25.2 >23.4 >24.9 >24.4 >25.3 >20.8 >22.2 >21.5 >21.5 >21.3 >21.2 >22.1 >22.2 >23.5 >23.0 >23.2 >23.7 >26.1 >26.5 >27.1 >26.2 >26.6 >26.0 >23.4 >22.6 >23.0 >23.1 >22.9 >22.5 >26.7 >27.3 >27.8 >28.4 >26.0 >26.0 >27.0 >26.4
76.9 79.5 46.2 92.3 42.3 80.8 26.9 61.5 85.6 88.6 92.3 61.5 76.9 76.9 88.5 69.2 72.1 71.8 76.9 80.8 87.2 68.7 57.7 80.8 82.1 65.4 84.6 88.5 78.8 88.5 69.2 87.2 65.4 73.1 57.7 61.5 58.3 66.7 83.3 58.3 57.7 65.4 84.6 80.8 61.5 69.2
62.8 60.9 77.0 27.3 73.9 42.5 84.3 62.7 28.5 50.4 40.4 80.8 72.4 63.1 38.8 54.7 62.0 52.6 43.6 29.3 56.4 56.7 56.4 36.0 60.9 61.1 34.2 46.0 65.3 47.7 55.6 45.0 49.7 41.2 56.9 58.8 56.3 59.2 40.8 53.5 53.2 58.4 33.1 38.3 52.6 48.7
0.691 0.688 0.616 0.598 0.582 0.629 0.540 0.614 0.527 0.645 0.686 0.723 0.782 0.688 0.624 0.640 0.725 0.733 0.602 0.593 0.763 0.714 0.563 0.583 0.675 0.644 0.582 0.638 0.639 0.668 0.595 0.659 0.559 0.570 0.588 0.595 0.580 0.572 0.609 0.574 0.622 0.661 0.575 0.546 0.611 0.553
1.23 1.12 0.85 0.72 0.56 0.86 0.33 0.61 0.20 1.02 1.17 1.14 1.27 1.16 0.86 0.99 1.11 1.25 0.79 0.73 1.14 1.22 0.56 0.66 1.24 0.96 0.65 1.06 0.99 1.17 0.52 1.09 0.43 0.76 0.81 0.95 0.92 1.03 1.58 1.16 0.46 0.64 0.20 0.05 0.35 0.06
.007 .024 .117 .203 .282 .140 .496 .281 .757 .143 .033 .018 .004 .012 .086 .036 .020 .019 .107 .103 .040 .037 .277 .181 .024 .087 .251 .060 .119 .062 .214 .015 .426 .162 .151 .337 .315 .351 .226 .324 .146 .038 .563 .897 .262 .850
AUC, area under the curve; ROI, region of interest; SR, success rate; DT is the difference in mean adjusted temperature, group 2–group 1. Bold values indicate the body regions with statistically significant differences in adjusted ROI temperature between group 1 and group 2.
used to identify developing injury which might cause an elevation in temperature not just suggestive of increased load because of high performance but subclinical inflammation due to training overload. In our simple ROC analysis, AUC values were low, and none of the ROIs, considered in isolation, would be able to classify an individual racehorse as a poor or successful performer reliably. More analysis is now required to consider if a combination of ROIs could provide better discriminatory power between groups. In the limbs, the best ROIs for discriminating performance were identified by our analysis as the carpal joints and the third metacarpal bones on both sides of the body.
In contrast, we found ROIs at the fetlock joint, front short pastern bone, and tarsus joint to be of utility only on the left side. The reason for this is unclear, and further investigation is required to establish how these findings might relate to equine biomechanics. Possibly, as horses during racing and exercise were loaded on the right front limb because of the clockwise race track, when resting, they were overloading the opposite side. 5. Conclusions The study tested a protocol for recording body surface temperature in racehorses, which was shown to increase
M. Soroko et al. / Journal of Equine Veterinary Science 34 (2014) 1076–1083
reliability by adjusting for variations in ambient temperature. Variation in temperature between sides of the body was generally <1 C at the defined measurement sites. When analyzed on the basis of sporting performance, the protocol identified 14 key ROIs that were associated with superior performance, most of which were at the limbs on the left side. Acknowledgments The authors thank Monika Słowik, MSc, the Director of Partynice Race Course, for kind permission to carry out the research on racehorses. The authors declare no conflicts of interest. References [1] Hamalin MJ, Shermann JP, Hopkins WG. Changes in physiological parameters in overtrained standardbred racehorses. Equine Vet J 2002;34:383–8. [2] Golland LC, Evans DL, Mcgowan CM, Hodgson DR, Rose RJ. The effects of overtraining on blood volumes in standardbred racehorses. Vet J 2003;165:228–33. M, Janczarek I. Współzalezno _ s [3] Kapron c mie˛ dzy wybranymi wska znikami zaawansowania treningowego i parametrami krwi koni wyscigowych. Zesz Nauk PTZ 2004;72:171–7. [4] Jodkowska E. Temperatura powierzchni ciała jako kryterium predyspozycji wysiłkowych konia. Zesz Nauk AR We Wrocławiu 2005;511. [5] Purohit RC, McCoy MD. Thermography in the diagnosis of inflammatory processes in the horse. Am J Vet Res 1980;41:1167–74. [6] Bowman KF, Purohit RC, Ganjam VK, Pechman Jr RD, Vaughan JT. Thermographic evaluation of corticosteroid efficacy in amphotericin B-induced arthritis in ponies. Am J Vet Res 1983;44:51–6. [7] Turner TA, Fessler JF, Lamp M, Pearce JA, Geddes LA. Thermographic evaluations of horses with podotrochlosis. Am J Vet Res 1983;44:535–9. [8] Turner TA. Thermography as an aid to the clinical lameness evaluation. Vet Clin N Am Equine Pract 1991;7:311–38. [9] Vaden MF, Purohit RC, Mc Coy D, Vaughan JT. Thermography: a technique for subclinical diagnosis of osteoarthritis. Am J Vet Res 1980;41:1175–9.
1083
[10] Turner TA, Pansch J, Wilson JH. Thermographic assessment of racing thoroughbreds. Proc Am Assoc Equine Pract 2001;47:344–6. [11] Turner TA. Back problems in horses. 49th Annual Convention of the Proc. Am. Assoc. Equine Pract., New Orleans, Louisiana 2003. [12] Turner TA. Diagnostic thermography. Vet Clin N Am Equine Pract 2001;17:95–113. [13] Tunley BV, Henson FM. Reliability and repeatability of thermographic examination and the normal thermographic image of the thoracolumbar region in the horse. Equine Vet J 2004;36:306–12. [14] Van Hoogmoed LM, Snyder JR, Allen AK, Waldsmith JD. Use of infrared thermography to detect performance-enhancing techniques in horses. Equine Vet Edu 2000;12:102–7. [15] Soroko M, Henklewski R, Filipowski H, Jodkowska E. The effectiveness of thermographic analysis in equine orthopedics. J Equine Vet Sci 2013;33:760–2. [16] Bruton A, Conway JH, Holgate ST. Reliability: what is it, and how is it measured? Physiotherapy 2000;86:94–9. [17] Egan JP. Signal detection theory and ROC analysis. Series in cognition and perception. New York: Academic Press; 1975. [18] Turner TA, Rantanen NW, Hauser ML. Alternate methods of soft tissue imaging. 1st Proc Dubai Inter Equine Symp 1996;165–176. [19] Von Schweinitz DG. Thermographic diagnosis in equine back pain. Vet Clin N Am Equine Pract 1999;15:161–77. [20] Fonseca BPA, Alves ALG, Nicoletti JLM, Thomassian A, Hussini CA, Mikaik S. Thermography and ultrasonography in back pain diagnosis of equine athletes. J Eq Vet Sci 2006;26:507–16. [21] Rossdale PD, Hopes R, Wingfield-Digby NJ, Offord K. Epidemiological study of wastage among racehorses 1982 and 1983. Vet Rec 1985;116:66–9. [22] De Cocq P, Weeren Van PR, Back W. Effects of girth, saddle and weight on movements of the horse. Equine Vet J 2004;36:758–63. [23] Schöllhorn WI, Peham C, Licka T, Scheidl M. A pattern recognition approach for the quantification of horse and rider interactions. Equine Vet J 2006;38:400–5. [24] Jeffcott LB, Rossdale PD, Freestone J, Frank CJ, Towers-Clark PF. An assessment of wastage in thoroughbred racing from conception to 4 years of age. Equine Vet 1982;14:185–98. [25] Williams RB, Harkins LS, Hammond CJ, Wood JL. Racehorses injuries, clinical problems and fatalities recorded on British racecourses from flat racing and national hunt racing during 1996, 1997 and 1998. Equine Vet J 2001;33:478–86. [26] Soroko M, Jodkowska E, Zabłocka M. The use of thermography to evaluate back musculoskeletal responses of young racehorses to training. Thermology Int 2012;22:152–6. [27] Soroko M. The effect of long-term training on racehorses’ body surface temperature changes. Phd dissertation, University of Environmental and Life Sciences of Wroclaw, Department of Horse Breeding and Equestrian Studies, 2013.
Contents lists available at ScienceDirect
Journal of Equine Veterinary Science journal homepage: www.j-evs.com
Original Research
Thermographic Evaluation of Racehorse Performance Maria Soroko PhD a, *, Krzysztof Dudek PhD b, Kevin Howell PhD c, Ewa Jodkowska a, Radomir Henklewski PhD d a
Department of Horse Breeding and Equestrian Studies, Wroclaw University of Environmental and Life Sciences, Wroclaw, Poland Institute of Machines Design and Operation, Technical University of Wroclaw, Wroclaw, Poland c Microvascular Diagnostics, Institute of Immunity and Transplantation, Royal Free Hospital, London, UK d Department of Surgery, Wroclaw University of Environmental and Life Sciences, Wroclaw, Poland b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 8 April 2014 Received in revised form 28 May 2014 Accepted 13 June 2014 Available online 25 June 2014
The present study was aimed at identifying the key thermographic diagnostic areas essential for monitoring the effect of training on racehorses. The study involved monitoring 15 racehorses in 13 imaging sessions over a period of 10 months. Temperature measurements were made at a total of 46 regions of interest (ROIs) at the distal parts of the limbs and the back. To account for the influence of ambient temperature on each ROI measurement, values were adjusted to a constant ambient temperature of 12 C, estimated using regression analysis. The horses in the study were divided into two groups based on the value of success rate in racing competition. During the research period, none of the horses were identified as injured by routine veterinary investigation. Successful horses had significantly warmer adjusted ROI temperatures than their less successful counterparts at both carpal joints, the third metacarpal bones, the left fetlock joint, the left front short pastern bone, the left tarsus joint, and the caudal part of the thoracic vertebrae. The study tested a protocol for recording body surface temperature in racehorses, which was shown to increase reliability by adjusting for variations in ambient temperature. When analyzed on the basis of sporting performance, the protocol identified 14 ROIs that were associated with superior performance, most of which were at the limbs on the left side. Ó 2014 Elsevier Inc. All rights reserved.
Keywords: Thermography Racehorse Sport performance Training
1. Introduction The health and training performance level of horses can be evaluated by behavioral, hematological, and biochemical changes in the body during physiological effort. Previous studies have investigated basic physiological indicators such as blood chemistry and pulse rate [1–3], but they have not provided evidence of the changes to the equine musculoskeletal system that occur in response to regular training. Jodkowska [4] identified the possibility of monitoring the impact of training through the measurement of body * Corresponding author at: Maria Soroko, PhD, Department of Horse Breeding and Equestrian Studies, Wroclaw University of Environmental and Life Sciences, Kozuchowska 5A, 51-161 Wroclaw, Poland. E-mail address: [email protected] (M. Soroko). 0737-0806/$ – see front matter Ó 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jevs.2014.06.009
surface temperature, which is influenced by the vascularity and metabolic activity within and below the skin surface. Thermography is a noninvasive measurement method enabling abnormal patterns of skin surface temperature to be detected by means of infrared imaging. Potential veterinary applications of thermography have been described [5–8]. These reports mostly address the detection of injuries caused by training overloads, especially at the back and the distal parts of the limbs of the horse [9–11]. Normal thermographic patterns in the horse have also been described, enabling improved diagnostic interpretation of equine thermography [5,12,13]. To date, however, no study has located the optimal body regions for reliable thermographic evaluation of musculoskeletal impact on horses. Therefore, the present study was aimed at identifying the key thermographic diagnostic areas essential for monitoring the effect of training on racehorses.
M. Soroko et al. / Journal of Equine Veterinary Science 34 (2014) 1076–1083
1077
2. Materials and Methods 2.1. Study Population and Data Collection The study was approved by the Local Ethical Committee for Experiments on Animals of Wroclaw University of Environmental and Life Sciences. Thermography measurements were obtained from 15 clinically healthy racehorses of two breeds (12 Polish Half Breed and three Arabians), all aged 3 years. All horses were trained for flat racing at Partynice Race Course (Poland) during the 2012 season. Performance data such as orthopedic injuries, training intensity, and success rate (SR, prize money per race) for 2012 were collected for each horse. The horses were housed in individual stalls with common management and training regimes. All horses were trained and raced in a clockwise direction. The protocol for thermography was as previously described by Van Hoogmoed et al [14] and Soroko et al [15]. Horses were examined at rest before daily exercise. Dirt and mud present in the imaging field of view was brushed away 15 minutes before examination and approximately 10 minutes were allowed to pass before scanning to ensure the transient heat generated by brushing had subsided before obtaining baseline measurements. The brushing was performed symmetrically. To minimize the effect of environmental factors, thermography was always performed at the same place within an enclosed stable. The distance of the animal from the camera was fixed for all imaging at 1 m, and the emissivity (3) was set to 1 for all readings. At each imaging session, a series of six thermographic images were taken of the dorsal, lateral, and medial aspects of the distal part of the forelimbs (Figs. 1–3); the dorsal aspect of the distal part of the hind limbs (Figs. 4 and 5); and the dorsal aspect of the back (Fig. 6) using a VarioCam HR infrared camera (uncooled microbolometer focal plane array; resolution, 640 480 pixels; spectral range,
7.5–14 mm; InfraTec, Dresden, Germany). A total of 13 imaging sessions were conducted over a period of 10 months. At each session, the ambient temperature in the stable (Tamb) was measured by a TES 1314 thermometer (TES, Taipei, Taiwan). Regions of interest (ROIs) were determined for each thermographic image. On each image of the forelimbs 10 ROIs were defined (Figs. 1–3), allowing temperature measurements to be made bilaterally at the dorsal, lateral, and medial aspects of the carpal joint, third metacarpal bone, fetlock joint and short pastern bone; the dorsal aspect of the hoof; and the lateral and medial aspects of the heel. On each image of the hind limbs 5 ROIs were defined (Figs. 4 and 5) allowing bilateral measurements at the
Fig. 1. Thermogram of dorsal aspect of distal part of forelimbs. Measured regions of interest: DF1dright carpal joint, dorsal aspect; DF2dleft carpal joint, dorsal aspect; DF3dright third metacarpal bone, dorsal aspect; DF4dleft third metacarpal bone, dorsal aspect; DF5dright fetlock joint, dorsal aspect; DF6dleft fetlock joint, dorsal aspect; DF7dright short pastern bone, dorsal aspect; DF8dleft short pastern bone, dorsal aspect; DF9dright hoof, dorsal aspect; and DF10dleft hoof, dorsal aspect.
Fig. 3. Thermogram of left lateral and right medial aspects of distal part of the forelimbs. Measured regions of interest: MF1dright carpal joint, medial aspect; LF2dleft carpal joint, lateral aspect; MF3dright third metacarpal bone, medial aspect; LF4dleft third metacarpal bone, lateral aspect; MF5dright fetlock joint, medial aspect; LF6dleft fetlock joint, lateral aspect; MF7dright short pastern bone, medial aspect; LF8dleft short pastern bone, lateral aspect; MF9dright heel, medial aspect; and LF10dleft heel, lateral aspect.
Fig. 2. Thermogram of right lateral and left medial aspects of distal part of the forelimbs. Measured regions of interest: LF1dright carpal joint, lateral aspect; MF2dleft carpal joint, medial aspect; LF3dright third metacarpal bone, lateral aspect; MF4dleft third metacarpal bone, medial aspect; LF5dright fetlock joint, lateral aspect; MF6dleft fetlock joint, medial aspect; LF7dright short pastern bone, lateral aspect; MF8dleft short pastern bone, medial aspect; LF9dright heel, lateral aspect; and MF10dleft heel, medial aspect.
1078
M. Soroko et al. / Journal of Equine Veterinary Science 34 (2014) 1076–1083
Fig. 4. Thermogram of dorsal aspect of distal part of right hind limb. Measured regions of interest: DH1dright tarsus joint, dorsal aspect; DH3dright third metatarsal bone; dorsal aspect; DH5dright hind fetlock joint dorsal aspect; DH7dright hind short pastern bone, dorsal aspect; and DH9dright hind hoof, dorsal aspect.
dorsal aspect of the tarsus joint, the third metatarsal bone, fetlock joint, short pastern bone, and hoof. On the image of the spine 6 ROIs were defined (Fig. 6) to obtain measurements from the dorsal aspect of the cranial and caudal parts of the thoracic vertebrae; the left and right sides of the thoracic vertebrae; the lumbar vertebrae; and the sacroiliac joints. The temperature was calculated using IRBIS 3 Professional software (InfraTec, Dresden, Germany) using the mean pixel value in the circular ROI for joints, hoofs, and heels; the mean pixel value along the linear ROI for long bones; and the mean value within the polygon for ROIs at the back. The average temperature was obtained from 8,786 ROIs. Not all ROIs were always available for analysis
Fig. 6. Thermogram of dorsal aspect of back. Measured regions of interest: B1dcranial part of thoracic vertebrae, dorsal aspect; B2dcaudal part of thoracic vertebrae, dorsal aspect; BL1dleft side of thoracic vertebrae, dorsal aspect; BR1dright side of thoracic vertebrae, dorsal aspect; B3dlumbar vertebrae, dorsal aspect; and B4dsacroiliac joints.
due to premature withdrawal of the horse from the racing season or inaccessibility of the body region for imaging. For estimation of repeatability of the analysis, ROIs were placed twice by a single operator on the thermographic images collected in the same session in 10 randomly chosen horses. Every 3 months, standard veterinary methods were used for palpation examinations of the distal parts of the forelimbs, hind limbs and back. Additional ultrasonographic and radiographic examinations were performed for the forelimbs. These were conducted by a veterinarian to diagnose any pathologic conditions. For ultrasonographic examinations, an Echo Blaster 128 scanner (Telemed, Vilnius, Lithuania) was used, and for radiographic examinations, Gierth HF90 (Gierth, Riesa, Germany) and Conaxx 35 (PROTEC, Oberstenfeld, Germany) systems were used. 2.2. Statistical Analysis All statistical analysis was performed using STATISTICA v. 10 (StatSoft, Tulsa). 2.2.1. ROI Temperature Adjustment At each body site, the correlation of ROI temperature (TROI) with ambient temperature (Tamb) was analyzed. To account for the influence of ambient temperature on each ROI temperature measurement, TROI values were adjusted adj to a value TROI for a constant ambient temperature of 12 C, estimated using the equation:
adj 2 TROI ¼ TROI þ b1 ,ð12 Tamb Þ þ b2 , 122 Tamb Model coefficients, b1 and b2, were estimated for each ROI separately using regression analysis. An example of adjustment for two readings taken from the ROI DF1 is given in Fig. 7. Fig. 5. Thermogram of dorsal aspect of distal part of left hind limb. Measured regions of interest: DH2dleft tarsus joint, dorsal aspect; DH4dleft third metatarsal bone, dorsal aspect; DH6dleft hind fetlock joint, dorsal aspect; DH8dleft hind short pastern bone, dorsal aspect; and DH10dleft hind hoof, dorsal aspect.
2.2.2. Coefficient of Variation in Temperature Within Each ROI The range of pixel temperature values within an ROI depends on its location on the body of a horse (related to the musculature, vasculature, fat thickness, and so forth)
M. Soroko et al. / Journal of Equine Veterinary Science 34 (2014) 1076–1083
1079
(SR): group 1: SR V400/race, 12 horses (including eight mares) and 152 pooled measurement results and group 2: SR > V400/race, three horses (including two mares) and 39 pooled measurement results. The difference in mean adjusted temperature (DT) between group 2 horses (SR > V400) and group 1 horses (SR V400) was calculated for each ROI. Statistical significance was tested using Student t test, with results considered significant if P < .05. An example box-andwhisker plot for group 1 and group 2 horses is shown for ROI DF1 in Fig. 8.
Fig. 7. Example of temperature adjustment principle to ambient temperature Tamb ¼ 12 C for two readings at the region of interest DF1dright carpal joint, dorsal aspect. Tamb, ambient temperature; TDF1, unadjusted temperature of ROI DF1; TadjDF1, temperature of ROI DF1 adjusted to an ambient temperature of 12 C.
and also on the size of the surface area. A uniform temperature across the ROI will assist in improved repeatability of results and implies that the ROI size and position is physiologically relevant. The coefficient of variation (CV) was adopted as the benchmark of temperature uniformity within each ROI:
CV ¼
SD ,100; M
where SD is the standard deviation of the pixel temperature values and M is the mean temperature value within the ROI, both extracted from the IRBIS 3 Professional software output. The mean values of the CV (across all horses and all measurement sessions) are presented in Table 1. 2.2.3. Repeatability of the Results Expressed by Intraclass Correlation Coefficient rICC Measurement repeatability in each of the 46 ROI was based on intraclass correlation coefficients, rICC. The following formula was used [16]:
rICC ¼
MSBS MSres MSBS þ ðk 1Þ,MSres þ nk ðMSBC MSres Þ
where MSBCdmean squares between couples; MSBSd mean of squares between subjects; MSresdmean squared residual n–number of horses (n ¼ 10); kdnumber of measurements (k ¼ 2). The value of rICC z 1 indicates a strong absolute temperature consistency in the assessment of ROI by the evaluator. This is reflected in a large variance between the objects (a significant difference between average n measurements) and low variance among repeated measurements by the evaluator (an insignificant difference in average results among k repeated measurements). 2.2.4. Difference in Mean ROI Temperature Among Horses With Different Success Rates The average prize won by horses competing in the 2012 season was V400/race. The horses in the study were divided into two groups based on this value of success rate
2.2.5. Area Under the Receiver Operating Characteristic Curve Receiver operating characteristic (ROC) curve analysis [17] makes it possible to identify whether the test parameter (adjusted temperature) of the ROI allows the correct classification of the analyzed horse to one of the two groups (group 1, SR V400 or group 2, SR > V400) in the most reliable way. Receiver operating characteristic analysis identifies a “cutoff” temperature which optimizes the sensitivity and specificity of the test for classifying the horse into the correct group. An example for ROI LF2 is given in Fig. 9, wherein (100dspecificity) is plotted against sensitivity. A cutoff at the top-left corner of the plot would be able to group the analyzed horse with perfect sensitivity and specificity. In this case, the area under the curve (AUC) would be 1. Conversely, a cutoff lying along the dotted line of identity would have no discriminatory power for grouping the analyzed horse, and in this case, the AUC would be 0.5. The optimum cutoff for region LF2 is marked along the ROC curve with a dot, and the AUC is 0.782. 3. Results During the research period, none of the horses were identified as injured by the ultrasonographic, radiographic, or palpation examinations. The ambient temperature in the stable Tamb ranged between 2.8 C and 22.7 C (Fig. 7). Temperature data derived from the 46 ROIs are presented in Table 1. The table shows mean and standard deviation ROI temperature values ðT ROI SDÞ (all horses, all measurement sessions); minimum temperature recorded for the ROI (tmin) and the maximum value (tmax); mean CV of pixel temperature values within each ROI; intraclass correlation coefficients (rICC) for the repeated readings from 10 horses; correlation coefficient (R) of ROI temperature with Tamb and the values of the regression coefficients b1 and b2; mean adjusted ROI temperatures to a constant ambient temperadj ature of 12 C (all horses, all measurement sessions) ðT ROI Þ. The mean CV for pixel values within each ROI was generally low, with the highest mean CV being 6.4% at ROI LF6 (left fetlock joint, lateral aspect). Intraclass correlation coefficients (rICC) for the repeated readings from 10 horses were generally high: 38/46 ROIs had an rICC greater than 0.8. The lowest repeatability was observed at ROIs MF1 (right carpal joint, medial aspect), rICC ¼ 0.614, and MF2 (left carpal joint, medial aspect), rICC ¼ 0.626. Adjustment of the ROI temperatures using the regression analysis limited the influence of ambient temperature on each mean ROI reading. Standard deviations on the adjusted mean temperature readings were greatly reduced
1080
M. Soroko et al. / Journal of Equine Veterinary Science 34 (2014) 1076–1083
Table 1 Basic temperature statistics of analyzed regions of interest (ROIs). Region (ROI)
T ROI SD
DF1dright carpal joint, dorsal aspect DF2dleft carpal joint, dorsal aspect DF3dright third metacarpal bone, dorsal aspect DF4dleft third metacarpal bone, dorsal aspect DF5dright fetlock joint, dorsal aspect DF6dleft fetlock joint, dorsal aspect DF7dright short pastern bone, dorsal aspect DF8dleft short pastern bone, dorsal aspect DF9dright hoof, dorsal aspect DF10dleft hoof, dorsal aspect LF1dright carpal joint, lateral aspect MF1dright carpal joint, medial aspect LF2dleft carpal joint, lateral aspect MF2dleft carpal joint, medial aspect LF3dright third metacarpal bone, lateral aspect MF3dright third metacarpal bone, medial aspect LF4dleft third metacarpal bone, lateral aspect MF4dleft third metacarpal bone, medial aspect LF5dright fetlock joint, lateral aspect MF5dright fetlock joint, medial aspect LF6dleft fetlock joint, lateral aspect MF6dleft fetlock joint, medial aspect LF7dright short pastern bone, lateral aspect MF7dright short pastern bone, medial aspect LF8dleft short pastern bone, lateral aspect MF8dleft short pastern bone, medial aspect LF9dright heel, lateral aspect MF9dright heel, medial aspect LF10dleft heel, lateral aspect MF10dleft heel, medial aspect DH1dright tarsus joint, dorsal aspect DH2dleft tarsus joint, dorsal aspect DH3dright third metatarsal bone, dorsal aspect DH4dleft third metatarsal bone, dorsal aspect DH5dright hind fetlock joint, dorsal aspect DH6dleft hind fetlock joint, dorsal aspect DH7dright hind short pastern bone, dorsal aspect DH8dleft hind short pastern bone, dorsal aspect DH9dright hind hoof, dorsal aspect DH10dleft hind right hoof, dorsal aspect B1dcranial part of thoracic vertebrae, dorsal aspect B2dcaudal part of thoracic vertebrae, dorsal aspect BL1dleft side of thoracic vertebrae, dorsal aspect BR1dright side of thoracic vertebrae, dorsal aspect B3dlumbar vertebrae, dorsal aspect B4dsacroiliac joints, dorsal aspect
25.5 25.3 24.0 24.2 24.4 24.6 24.0 24.1 27.7 27.6 25.4 25.8 26.4 26.3 24.2 24.3 24.4 24.3 24.4 24.4 24.9 24.6 25.8 25.7 26.1 25.8 28.7 28.4 28.5 28.3 28.3 28.2 25.6 25.5 24.9 24.8 21.2 21.1 25.7 25.8 29.4 29.9 28.7 28.5 28.9 28.5
5.1 5.1 5.5 5.6 5.3 5.4 5.2 5.2 4.9 4.9 5.5 5.1 5.0 4.9 5.7 5.7 5.7 5.7 5.6 5.5 5.9 5.9 5.3 5.1 5.4 5.4 4.7 4.8 4.8 4.7 4.5 4.6 5.3 5.3 5.2 5.2 4.9 4.8 5.2 4.8 4.5 4.3 5.3 5.3 4.7 5.0
adj
tmin
tmax
Mean CV (%)
rICC
R
b1
b2
T ROI SD
9.9 10.5 7.6 7.8 8.1 7.7 7.9 7.2 9.1 9.0 11.7 12.1 12.6 12.1 8.4 9.1 8.4 8.3 7.6 8.1 9.0 8.0 9.6 9.2 8.8 8.2 10.6 10.8 9.1 9.4 12.5 14.5 8.6 10.2 7.2 9.6 6.7 7.9 10.4 11.4 18.4 19.1 15.2 14.3 16.0 16.4
32.5 32.6 32.6 32.8 32.5 32.4 31.4 31.4 34.0 33.9 33.5 34.1 33.5 34.2 32.4 32.4 32.6 32.6 32.5 31.7 32.4 32.5 32.9 33.0 33.5 33.2 35.0 34.7 34.8 34.4 34.2 33.7 32.5 32.4 31.8 31.8 29.4 29.9 32.5 32.5 35.1 35.4 35.1 34.8 35.0 34.9
2.8 2.8 2.3 2.4 2.9 3.2 3.3 3.2 2.8 3.1 4.3 5.6 3.6 5.8 4.8 5.5 4.6 4.9 6.3 5.9 6.4 5.8 4.4 5.3 4.1 5.7 5.2 5.4 5.0 5.0 3.0 3.2 5.5 2.9 3.1 5.1 5.5 5.9 4.3 4.2 2.3 2.1 3.0 2.9 1.6 1.6
0.920 0.937 0.981 0.960 0.951 0.943 0.940 0.913 0.942 0.922 0.810 0.614 0.682 0.626 0.892 0.861 0.949 0.959 0.896 0.878 0.882 0.868 0.965 0.923 0.934 0.949 0.878 0.809 0.815 0.881 0.920 0.923 0.860 0.898 0.955 0.954 0.777 0.789 0.722 0.825 0.849 0.803 0.773 0.770 0.970 0.985
0.805 0.789 0.788 0.772 0.788 0.741 0.805 0.746 0.630 0.566 0.845 0.847 0.819 0.828 0.847 0.844 0.839 0.828 0.856 0.852 0.801 0.805 0.812 0.798 0.767 0.784 0.704 0.694 0.613 0.624 0.809 0.790 0.774 0.749 0.759 0.746 0.593 0.584 0.351 0.394 0.882 0.878 0.896 0.889 0.897 0.900
0.84 0.82 0.93 0.92 0.89 0.86 0.77 0.73 0.52 0.38 0.87 0.94 0.87 0.45 0.46 1.02 1.04 0.43 0.44 0.96 1.02 0.47 0.44 0.88 0.82 0.43 0.40 0.78 0.33 0.62 1.08 1.10 1.01 1.03 0.90 0.97 0.68 0.64 0.40 0.35 0.99 1.00 1.07 1.10 1.27 1.27
0.001 0 0.002 0.002 0.002 0.001 0.002 0.003 0.006 0.010 0.002 0.006 0.003 0.010 0.014 0.004 0.005 0.015 0.015 0.003 0.004 0.013 0.013 0.002 0.001 0.012 0.009 0.003 0.011 0.001 0.012 0.013 0.006 0.006 0.004 0.006 0.002 0.004 0.006 0.010 0.009 0.011 0.010 0.010 0.015 0.015
23.0 23.0 21.7 21.8 22.1 22.3 21.7 21.8 25.6 25.5 24.4 24.6 24.8 25.0 21.8 22.0 22.0 21.8 21.9 22.0 22.3 21.9 23.2 23.6 23.3 23.0 26.8 26.5 28.0 28.2 26.5 26.4 22.5 22.7 22.7 22.2 22.1 22.1 26.4 26.4 27.9 28.2 26.7 26.6 27.0 26.8
2.3 2.3 2.6 2.7 2.4 2.7 2.3 2.6 3.0 3.3 2.1 2.0 2.1 2.2 2.4 2.2 2.3 2.5 2.3 2.1 2.6 2.8 2.4 2.3 2.6 2.6 3.1 2.7 3.0 3.0 2.0 2.1 2.5 2.5 2.7 3.2 3.1 3.0 4.2 3.7 1.5 1.4 1.6 1.7 1.5 1.5
b1, b2, binomial regression coefficients of ROI with Tamb; Mean CV, mean coefficient of variation of pixel temperature values for each ROI; R, correlation coefficient of ROI temperature with ambient temperature (Tamb); rICC, intraclass correlation coefficient for repeatability of readings (two repeat readings, 10 horses); tmax, maximum temperature value ( C); tmin, minimum temperature value ( C); T ROI SD, mean and standard deviation of temperature in ROI ( C); adj T ROI SD, mean and standard deviation of adjusted temperature in ROI ( C).
compared with SDs for the unadjusted data. Adjusted ROI temperatures showed a generally high degree of symmetry between the left and right sides. Asymmetry for all sites was less than 1 C, with the exception of the lateral and medial aspects of the front heels. The ROC and Student t test analysis of SR is detailed in Table 2. Successful (group 2) horses had warmer adjusted ROI temperatures than their less successful counterparts (group 1) at all body sites. Differences in adjusted temperature between the groups were significantly higher (P < .05) in 14 of 46 sites (bold values in the table), including all aspects of both carpal joints, some views of the third metacarpal bones, the lateral and medial aspects of the left fetlock joint, the lateral aspect of the left front short
pastern bone, the dorsal aspect of the left tarsus joint, and the caudal part of the thoracic vertebrae. The highest AUC values were registered in the adj following regions: TLF2 dleft carpus joint, medial aspect adj dleft fetlock joint, lateral aspect (AUC ¼ 0.782); TLF6 adj dleft third metacarpal bone, medial (AUC ¼ 0.763); and TLF4 aspect (AUC ¼ 0.733).
4. Discussion Thermography has been most commonly used to evaluate horses with limb lameness. Clinical diseases that have been recognized and characterized by thermography
M. Soroko et al. / Journal of Equine Veterinary Science 34 (2014) 1076–1083
Fig. 8. Example of comparison of adjusted temperatures of horses from groups 1 and 2 for region of interest DF1dright carpal joint, dorsal aspect. SE, standard error; SR, success rate; TadjDF1, temperature of ROI DF1 adjusted to an ambient temperature of 12 C.
include tendonitis, laminitis, and sole abscesses. It has also been successfully used for back abnormality diagnosis associated with neuromuscular disease of the thoracolumbar region and inflammation of spinous processes [8,18–20]. Most racehorse injures are associated with the physical demands on the musculoskeletal system in response to training [21], and they can also be linked to the type of training and the skills of the rider [22,23]. The detection and monitoring of physiological responses to training overloads are particularly important for racehorses put under extreme physical demands. In the study presented by Jeffcott et al [24], of 163 Thoroughbred racehorses, 53% suffered from lameness, which was the main cause of elimination of the horses from sport performance. A more recent study involving Thoroughbred racehorses recorded an 81% incidence of lameness [25]. Thermography has been used to identify areas of inflammation that could decrease racehorse performance [26]. In another study, it was found that the temperature of the back and the distal parts of the forelimbs in racehorses at rest was increased by long-term training [27]. In the present study, 10 months of regular thermographic examination of racehorses allowed the identification
1081
of the most important body regions to be recommended for monitoring the impact of training. Identifying the key thermographic diagnostic areas should facilitate the detection of pathologic conditions during the training cycle. This is particularly important in racehorses, where immediate diagnosis might help to maintain their health and condition, positively influencing their further career in sport. The determination of the ROIs we used was based on thermographic measurement standards in veterinary medicine. Repeatability of the thermographic measurements was generally good, with the notable exception of the medial aspect of both carpal joints. This poor repeatability may arise because these ROIs must be imaged at an angle around the contralateral limb, which might limit reproducibility of the imaging technique. We introduced an adjustment to take into consideration the effect of ambient temperature on body surface temperature; this reduced variability in the measurements and allowed ROIs that conveyed the best information about racehorse performance to be more reliably identified. We observed no correlation (P > .05) between the ambient temperature at which the horses were studied and SR. Hence the process of adjusting ROIs for ambient temperature was unlikely to systematically influence the study’s ability to classify horse performance by analyzing ROI temperatures. Our study has a number of important limitations. First, we were not able to collect data at consistent time points after each race, and this may reduce the ability of the study to classify the horses’ performance on the basis of ROI temperature. Second, we used the prize money earned by each horse as a measure of performance. Prize earnings are of course influenced not just by horse performance but by many additional factors such as the jockey and the standard of competition. Third, the number of horses in our study was small, with a bias toward poorer performing horses. A larger study with an even distribution of performance would increase the validity of the thermographic findings. None of the horses were injured during the measurement period, but nonetheless, the best performing racehorses (group 2) were warmer than their poorer performing peers (group 1) at all body sites studied. This temperature difference may reflect an increased physiological training impact on the best performing horses. Further work is required to establish if our protocol can be
adj Fig. 9. Receiver operating characteristic curve for LF2dleft carpal joint, lateral aspect; cutoff point for TLF2 > 24.4 C; and histogram for groups 2 and 1. AUC, area under the curve; ROI, region of interest; Sens, sensitivity; Spec, specificity; Tadj, temperature of ROI adjusted to an ambient temperature of 12 C.
1082
M. Soroko et al. / Journal of Equine Veterinary Science 34 (2014) 1076–1083
Table 2 Receiver operating characteristic (ROC) analysis and Student t test results for discrimination of horses in group 2 (SR > V400) from those in group 1 (SR V400). Region (ROI)
DF1dright carpal joint, dorsal aspect DF2dleft carpal joint, dorsal aspect DF3dright third metacarpal bone, dorsal aspect DF4dleft third metacarpal bone, dorsal aspect DF5dright fetlock joint, dorsal aspect DF6dleft fetlock joint, dorsal aspect DF7dright short pastern bone, dorsal aspect DF8dleft short pastern bone, dorsal aspect DF9dright hoof, dorsal aspect DF10dleft hoof, dorsal aspect LF1dright carpal joint, lateral aspect MF1dright carpal joint, medial aspect LF2dleft carpal joint, lateral aspect MF2dleft carpal joint, medial aspect LF3dright third metacarpal bone, lateral aspect MF3dright third metacarpal bone, medial aspect, LF4dleft third metacarpal bone, lateral aspect MF4dleft third metacarpal bone, medial aspect LF5dright fetlock joint, lateral aspect MF5dright fetlock joint, medial aspect LF6dleft fetlock joint, lateral aspect MF6dleft fetlock joint, medial aspect LF7dright short pastern bone, lateral aspect MF7dright short pastern bone, medial aspect LF8dleft short pastern bone, lateral aspect MF8dleft short pastern bone, medial aspect LF9dright heel, lateral aspect MF9dright heel, medial aspect LF10dleft heel, lateral aspect MF10dleft heel, medial aspect DH1dright tarsus joint, dorsal aspect DH2dleft tarsus joint, dorsal aspect DH3dright third metatarsal bone, dorsal aspect DH4dleft third metatarsal bone, dorsal aspect DH5dright hind fetlock joint, dorsal aspect DH6dleft hind fetlock joint, dorsal aspect DH7dright hind short pastern bone, dorsal aspect DH8dleft hind short pastern bone, dorsal aspect DH9dright hind hoof, dorsal aspect DH10dleft hind right hoof, dorsal aspect B1dcranial part of thoracic vertebrae, dorsal aspect B2dcaudal part of thoracic vertebrae, dorsal aspect BL1dleft side of thoracic vertebrae, dorsal aspect BR1dright side of thoracic vertebrae, dorsal aspect B3dlumbar vertebrae, dorsal aspect B4dsacroiliac joints, dorsal aspect
ROC analysis
Group 2 Versus Group 1
Cutoff ( C)
Sensitivity (%)
Specificity (%)
AUC
DT ( C)
P
>23.4 >23.4 >23.2 >22.4 >23.2 >22.9 >23.6 >22.5 >24.9 >25.2 >23.4 >24.9 >24.4 >25.3 >20.8 >22.2 >21.5 >21.5 >21.3 >21.2 >22.1 >22.2 >23.5 >23.0 >23.2 >23.7 >26.1 >26.5 >27.1 >26.2 >26.6 >26.0 >23.4 >22.6 >23.0 >23.1 >22.9 >22.5 >26.7 >27.3 >27.8 >28.4 >26.0 >26.0 >27.0 >26.4
76.9 79.5 46.2 92.3 42.3 80.8 26.9 61.5 85.6 88.6 92.3 61.5 76.9 76.9 88.5 69.2 72.1 71.8 76.9 80.8 87.2 68.7 57.7 80.8 82.1 65.4 84.6 88.5 78.8 88.5 69.2 87.2 65.4 73.1 57.7 61.5 58.3 66.7 83.3 58.3 57.7 65.4 84.6 80.8 61.5 69.2
62.8 60.9 77.0 27.3 73.9 42.5 84.3 62.7 28.5 50.4 40.4 80.8 72.4 63.1 38.8 54.7 62.0 52.6 43.6 29.3 56.4 56.7 56.4 36.0 60.9 61.1 34.2 46.0 65.3 47.7 55.6 45.0 49.7 41.2 56.9 58.8 56.3 59.2 40.8 53.5 53.2 58.4 33.1 38.3 52.6 48.7
0.691 0.688 0.616 0.598 0.582 0.629 0.540 0.614 0.527 0.645 0.686 0.723 0.782 0.688 0.624 0.640 0.725 0.733 0.602 0.593 0.763 0.714 0.563 0.583 0.675 0.644 0.582 0.638 0.639 0.668 0.595 0.659 0.559 0.570 0.588 0.595 0.580 0.572 0.609 0.574 0.622 0.661 0.575 0.546 0.611 0.553
1.23 1.12 0.85 0.72 0.56 0.86 0.33 0.61 0.20 1.02 1.17 1.14 1.27 1.16 0.86 0.99 1.11 1.25 0.79 0.73 1.14 1.22 0.56 0.66 1.24 0.96 0.65 1.06 0.99 1.17 0.52 1.09 0.43 0.76 0.81 0.95 0.92 1.03 1.58 1.16 0.46 0.64 0.20 0.05 0.35 0.06
.007 .024 .117 .203 .282 .140 .496 .281 .757 .143 .033 .018 .004 .012 .086 .036 .020 .019 .107 .103 .040 .037 .277 .181 .024 .087 .251 .060 .119 .062 .214 .015 .426 .162 .151 .337 .315 .351 .226 .324 .146 .038 .563 .897 .262 .850
AUC, area under the curve; ROI, region of interest; SR, success rate; DT is the difference in mean adjusted temperature, group 2–group 1. Bold values indicate the body regions with statistically significant differences in adjusted ROI temperature between group 1 and group 2.
used to identify developing injury which might cause an elevation in temperature not just suggestive of increased load because of high performance but subclinical inflammation due to training overload. In our simple ROC analysis, AUC values were low, and none of the ROIs, considered in isolation, would be able to classify an individual racehorse as a poor or successful performer reliably. More analysis is now required to consider if a combination of ROIs could provide better discriminatory power between groups. In the limbs, the best ROIs for discriminating performance were identified by our analysis as the carpal joints and the third metacarpal bones on both sides of the body.
In contrast, we found ROIs at the fetlock joint, front short pastern bone, and tarsus joint to be of utility only on the left side. The reason for this is unclear, and further investigation is required to establish how these findings might relate to equine biomechanics. Possibly, as horses during racing and exercise were loaded on the right front limb because of the clockwise race track, when resting, they were overloading the opposite side. 5. Conclusions The study tested a protocol for recording body surface temperature in racehorses, which was shown to increase
M. Soroko et al. / Journal of Equine Veterinary Science 34 (2014) 1076–1083
reliability by adjusting for variations in ambient temperature. Variation in temperature between sides of the body was generally <1 C at the defined measurement sites. When analyzed on the basis of sporting performance, the protocol identified 14 key ROIs that were associated with superior performance, most of which were at the limbs on the left side. Acknowledgments The authors thank Monika Słowik, MSc, the Director of Partynice Race Course, for kind permission to carry out the research on racehorses. The authors declare no conflicts of interest. References [1] Hamalin MJ, Shermann JP, Hopkins WG. Changes in physiological parameters in overtrained standardbred racehorses. Equine Vet J 2002;34:383–8. [2] Golland LC, Evans DL, Mcgowan CM, Hodgson DR, Rose RJ. The effects of overtraining on blood volumes in standardbred racehorses. Vet J 2003;165:228–33. M, Janczarek I. Współzalezno _ s [3] Kapron c mie˛ dzy wybranymi wska znikami zaawansowania treningowego i parametrami krwi koni wyscigowych. Zesz Nauk PTZ 2004;72:171–7. [4] Jodkowska E. Temperatura powierzchni ciała jako kryterium predyspozycji wysiłkowych konia. Zesz Nauk AR We Wrocławiu 2005;511. [5] Purohit RC, McCoy MD. Thermography in the diagnosis of inflammatory processes in the horse. Am J Vet Res 1980;41:1167–74. [6] Bowman KF, Purohit RC, Ganjam VK, Pechman Jr RD, Vaughan JT. Thermographic evaluation of corticosteroid efficacy in amphotericin B-induced arthritis in ponies. Am J Vet Res 1983;44:51–6. [7] Turner TA, Fessler JF, Lamp M, Pearce JA, Geddes LA. Thermographic evaluations of horses with podotrochlosis. Am J Vet Res 1983;44:535–9. [8] Turner TA. Thermography as an aid to the clinical lameness evaluation. Vet Clin N Am Equine Pract 1991;7:311–38. [9] Vaden MF, Purohit RC, Mc Coy D, Vaughan JT. Thermography: a technique for subclinical diagnosis of osteoarthritis. Am J Vet Res 1980;41:1175–9.
1083
[10] Turner TA, Pansch J, Wilson JH. Thermographic assessment of racing thoroughbreds. Proc Am Assoc Equine Pract 2001;47:344–6. [11] Turner TA. Back problems in horses. 49th Annual Convention of the Proc. Am. Assoc. Equine Pract., New Orleans, Louisiana 2003. [12] Turner TA. Diagnostic thermography. Vet Clin N Am Equine Pract 2001;17:95–113. [13] Tunley BV, Henson FM. Reliability and repeatability of thermographic examination and the normal thermographic image of the thoracolumbar region in the horse. Equine Vet J 2004;36:306–12. [14] Van Hoogmoed LM, Snyder JR, Allen AK, Waldsmith JD. Use of infrared thermography to detect performance-enhancing techniques in horses. Equine Vet Edu 2000;12:102–7. [15] Soroko M, Henklewski R, Filipowski H, Jodkowska E. The effectiveness of thermographic analysis in equine orthopedics. J Equine Vet Sci 2013;33:760–2. [16] Bruton A, Conway JH, Holgate ST. Reliability: what is it, and how is it measured? Physiotherapy 2000;86:94–9. [17] Egan JP. Signal detection theory and ROC analysis. Series in cognition and perception. New York: Academic Press; 1975. [18] Turner TA, Rantanen NW, Hauser ML. Alternate methods of soft tissue imaging. 1st Proc Dubai Inter Equine Symp 1996;165–176. [19] Von Schweinitz DG. Thermographic diagnosis in equine back pain. Vet Clin N Am Equine Pract 1999;15:161–77. [20] Fonseca BPA, Alves ALG, Nicoletti JLM, Thomassian A, Hussini CA, Mikaik S. Thermography and ultrasonography in back pain diagnosis of equine athletes. J Eq Vet Sci 2006;26:507–16. [21] Rossdale PD, Hopes R, Wingfield-Digby NJ, Offord K. Epidemiological study of wastage among racehorses 1982 and 1983. Vet Rec 1985;116:66–9. [22] De Cocq P, Weeren Van PR, Back W. Effects of girth, saddle and weight on movements of the horse. Equine Vet J 2004;36:758–63. [23] Schöllhorn WI, Peham C, Licka T, Scheidl M. A pattern recognition approach for the quantification of horse and rider interactions. Equine Vet J 2006;38:400–5. [24] Jeffcott LB, Rossdale PD, Freestone J, Frank CJ, Towers-Clark PF. An assessment of wastage in thoroughbred racing from conception to 4 years of age. Equine Vet 1982;14:185–98. [25] Williams RB, Harkins LS, Hammond CJ, Wood JL. Racehorses injuries, clinical problems and fatalities recorded on British racecourses from flat racing and national hunt racing during 1996, 1997 and 1998. Equine Vet J 2001;33:478–86. [26] Soroko M, Jodkowska E, Zabłocka M. The use of thermography to evaluate back musculoskeletal responses of young racehorses to training. Thermology Int 2012;22:152–6. [27] Soroko M. The effect of long-term training on racehorses’ body surface temperature changes. Phd dissertation, University of Environmental and Life Sciences of Wroclaw, Department of Horse Breeding and Equestrian Studies, 2013.