Since its beginnings in the second half of the 20th century, thermal imaging has reached a wide variety of applications. It is a non-contact way of measuring temperature, where a thermal imaging camera converts the thermal radiation of bodies into a thermographic image. This allows us to make both a qualitative and quantitative assessment of the observed physical phenomenon.
In this master's degree thesis, I investigate the applicability of thermal imaging for evaluation of racehorse performance. The main purpose is to determine the core body temperature and skin surface temperature reached by the horse during standard warm-up procedure and during race. I took measurements on nine trotter horses, measuring 6 different regions of interest, which were defined according to the muscle groups located under the skin. I also monitored the temperature dynamics of the regions after exertion and the temperature differences between each region of interest. Measurements of rectal temperature, heart rate and respiratory rate were also taken. Horses blood was sampled before standard warm-up procedure and after simulated race. From this samples hematological and biochemical parameters were determined. Three thermal imaging cameras of different price ranges were used to capture the thermograms to see how are results dependent on the quality of the thermal imaging camera. Thermograms were captured with FLIR T1020, FLIR T650sc and Fluke TiS45 cameras from different angles to cover all regions of interest. During the measurements, we monitored environmental parameters such as air temperature, air pressure, wind speed and relative humidity. These values were later used to analyse the thermograms correctly. FLIR camera thermograms were analysed with program FLIR ResearchIR, while Fluke TiS45 camera thermograms were analysed with Fluke Connect software. For the analysis, we observed mean value of temperatures, standard deviation of pixels and maximum temperature within individual regions of interest. Data was then analysed and differences between the individual camera and rectal temperature were compared. Thermograms with biggest measured differences were taken with the Fluke camera, which is in the low price range. While thermograms with the smallest differences were taken with the FLIR T1020, which is the most expensive camera and with the best specifications, in our study. I also performed a statistical analysis of the standard deviation of pixel temperatures for all regions of interest. Due to the small surface area of eye region, we get minimal standard deviations compared to other regions of interest, making the eye region an interesting area for further research.
I performed the correlation analysis for each horse separately, as a way to address the biological variability in population. I was looking for correlations between regions of interest and pulse, and rectal temperature. Since all data is not monotonous, I used Spearman's correlation coefficient and the corresponding p-test. After p-values we determines, outliers were subsequently removed using the Grubbs test. Three regions of interest had a very high correlation (0.80 - 1.00), of which all were located on the horse's torso. These were also the regions which had higher surface temperature dynamic. The anterior part of the torso has the strongest correlation with rectal temperature. This region could often be distinguished from the background even with the naked eye. For all three regions of interest, I used the differences between the regions and the rectal temperature to determine an equation, that can be used to calculate and approximation of the rectal temperature, from measurements made with thermal imaging camera.
I used blood count and biochemistry data to determine which laboratory parameters change significantly after physical exertion. Erythrocytes, haematocrit, haemoglobin, leucocytes and segmental leucocytes were significantly elevated (p < 0.05). Eosinophils and lymphocytes were significantly lower after fast training (p < 0.05). For biochemical parameters, muscle enzyme creatine kinase (p = 0.004) and sodium (p = 0.004) concentrations were elevated, while potassium (p = 0.038) was decreased.
The results confirm that there are correlations between skin surface temperature and clinical parameters. In practice, respiratory rate, heart rate and rectal temperature are the most commonly used parameters to assess the physiology of the horse. These parameters allow trainers to assess the fitness of the horse and whether it is at risk of hyperthermia. Thermal imaging is already used in the veterinary field, especially for early detection of diseases or injuries, before clinical symptoms appear. Further research could look in more detail at the regions of interest that were found statistically significant. With a larger population of horses and a more specific measurement protocol, a clearer measurement protocol could be defined that would be useful for monitoring the warm-up procedure of trotter horses.