The impact of the dairy cow’s position on eye and udder temperatures obtained with infra-red thermography within a walk-trough system

Abstract

Infra-red thermography (IRT) has the potential to detect disease, injury, and stress in dairy cows. Using IRT as a routine early warning system for such issues on-farm requires highly frequent imaging, which typically needs automation. To this end, automated systems collecting data every time an individual passes through a specified area have been developed (walk-through systems). Within these, the animal’s speed, exact path and posture affect its distance and angle from the camera. While such variation in positioning is known to impact recorded temperatures, the extent of this impact is rarely quantified. If the error due to suboptimal positioning is sufficient to obscure temperature changes associated with the condition to be detected, the reliability of an early warning system is greatly impaired. This study aimed to quantify the impact of positioning on IRT-derived temperatures. Multiple IRT images were obtained from 197 lactating Holstein-Frisians using a walk-trough system. We assessed specific body parts chosen for their practical relevance: the eye centre (used to detect stress or general ill-health) and udder and teat area (used to detect mastitis or teat stress). The location of the body part within the IRT image (“position category”) was used as a measure for combined changes in distance and angle of incidence. Minimisation of each of these two factors results in a maximisation of recorded temperature but was expected to occur at different position categories. Position category affected the recorded temperature of all three body parts on both sides of the cow (P < 0.0001). Temperatures peaked in position categories where distance was not yet fully minimised, underlining the importance of the angle of incidence. In images taken from the left side, recorded eye temperature showed a 2.2 °C (± 0.17 SEM) difference between the position where it peaked and the position where it bottomed out. This difference was 2.0 (± 0.07) and 1.5 (± 0.08) °C for maximum udder temperature and maximum teat area temperature, respectively. On the right side, these differences were 2.2 (± 0.28), 1.1 (± 0.11) and 0.6 (± 0.14) °C. The differences in temperature due to dairy cow positioning could mask the onset of a health problem, as these result in an approximately equal temperature rise. This suggests that even though walk-through systems standardise positioning to some extent, further standardisation is required. Our findings are not only of direct importance for the further development of walk-trough systems but also provide an insight into the optimisation of positioning when imaging freely moving animals.