Lack of agreement among analysers of infrared thermal images in the temperature of eye regions in sheep

Abstract

Infrared thermography (IRT) is a non-invasive method to estimate body temperature in many animal species. Theeye has been indicated as the ideal region for IRT, potentially serving as an early indicator of variations in bodytemperature triggered by various events and stimuli. Studies are being conducted to understand sources ofvariation in IRT temperature of the eye and develop comprehensive guidelines for veterinary medicine andanimal production purposes. The present study adds information about variation in temperature estimated fromdifferent eye regions in sheep due to analysers. Thirty-one thermal images, fifteen of the right and sixteen of theleft eyes of a ewe, captured with a FLIR E8-XT camera, were analysed by each of three analysers, and minimum(Tmin), maximum (Tmax) and average (Tavg) temperature in lacrimal caruncle (R1), medial canthus of the eye(R2) and eyeball (R3) were estimated. Differences among analysers were analysed by ANOVA (normal distri-bution) or Friedman test (non-normal distribution), considering thermal images as blocks. Limits of Agreementwith the Mean (LOAM) were estimated and agreement plots were drawn. There were no differences (P > 0.05)among analysers in temperature (◦C, mean ± standard deviation or median and 2.5%–97.5% quantiles) esti-mated in R1 (Tmin = 36.3 ± 0.87, Tmax = 37.2 (35.9–37.7), Tavg = 36.7 ± 0.67), R2 (Tmax = 37.2(36.0–37.8)) and R3 (Tavg = 36.1 (34.9–37.3)) for the right eye (n = 45), and in R1 (Tmin = 36.0 (34.3–37.3),Tmax = 36.8 ± 0.49, Tavg = 36.4 ± 0.53), R2 (Tmin = 35.9 ± 0.71, Tmax = 36.7 (35.8–37.7), Tavg = 36.3(35.5–37.5)) and R3 (Tavg = 35.9 (34.8–36.6)) for the left eye (n = 48); but differences (P < 0.05) were presentin R3 (Tmin, Tmax) for both eyes and in R2 (Tmin and Tavg) for the right eye. The highest mean differences intemperature (◦C) (P < 0.05) among analysers were in R3 from the right eye (Tmin = 0.70, Tmax = 0.25) and theleft eye (Tmin = 0.64, Tmax = 0.10), and in R2 from the right eye (Tmin = 0.54, Tavg = 0.50). Considering botheyes and all eye regions, the total range for the LOAM (◦C) was lower for Tmax (0.42–0.93) and Tavg (0.63–1.17)than for Tmin (1.49–2.70), indicating a higher, medium, and lower agreement among analysers for Tmax, Tavgand Tmin, respectively. Considering the eyes, the total range for the LOAM (◦C) was lower in all regions from theleft eye (0.42–1.59) compared to the right eye (0.79–2.70), indicating a higher agreement among analysers fromthe left eye. Considering the eye regions, the total range for the LOAM (◦C) was generally higher in R3(0.84–3.70) than in R1 (0.61–1.79) and R2 (0.42–2.03), indicating the lowest agreement among analysers in R3for Tmin (3.70). The present study shows that in sheep, the lacrimal caruncle confers the best agreement amonganalysers for all estimated temperatures, and that Tmax, across all eye regions, should be regarded as the mostreliable estimation.