Standardizing Infrared Thermography (IRT) and Micro-Behavioral Biometrics for Estrus Detection in Dairy Cows

Most Canadian dairy herds operate in tie-stall housing (61%) where estrus detection rates may be lower than 35%. Infrared thermography (IRT) is a non-invasive technology which can predict ovulation by measuring radiated heat. Complementary behavioral biometric algorithms may improve accuracy (>40%) and specificity (>6.7%). The objective of this study was to standardize infrared thermography and micro-behavioral biometrics for estrus detection in dairy cows. Thirty six cows were divided into 2 treatments with 18 pregnant (CON) and 18 open (OPEN) cows. Open cows were synchronized with GnRH and CIDR protocol on (-14 d) and 2 injections of PGF2α 12 hours apart on (-7 d). Pregnant cows received a sham injection and CIDR on same schedule and frequency as OPEN cows on the synchronization protocol. Cows were monitored via visual cameras (Swann DVR) 5 minutes before, during, and after milking to establish the frequency of treading, drinking, neighbor interactions, tail movements, laying and shifting behaviors. Radiated heat and physiological changes were monitored relative to estrus using an infrared camera (FLIR T450s). Infrared measurements were adjusted by recording environmental and relative humidity before and after thermogram collection. Infrared images were recorded for the eye, muzzle, cheeks, neck, front feet, round, heart girth, vulva surround, tail head, and withers and analysed using FLIR Tools software. Data were analysed using the Glimmix procedure in SAS (v9.4) with cow as the experimental unit. All results are reported as LSmeans ± SEM. Thermal biometrics differed by imaging location in CONT versus OPEN cows (eg. vulva: CON 34.8˚C ± 0.047; P<0.0001, OPEN 35.32˚C ± 0.047; P<0.0001; muzzle: CON 32.51˚C ± 0.09; P<0.0001, OPEN 33.45˚C ± 0.091; P<0.0001, and cheeks (CON 32.30˚C ± 0.084; P<0.0001, OPEN 31.22˚C ± 0.083; P<0.0001) Thermograms found OPEN cows cooled at the tail head and heart girth as ovulation approached (tail head: -3 d 33.2˚C ±0.2 vs. 0 d 32.7˚C ±0.23; P< 0.0367; heart girth: -3 d 33.21˚C ±0.19 vs. 0 d 32.66˚C ±0.22; P< 0.0097). Treading behavior was significantly higher in CON (20.84 ± 0.39; P<0.0001) and OPEN (16.35 ± 0.34; P<0.0001) cows. Tail movements increased 72 hours before ovulation (-3 d 9.95 ±1.18) compared with ovulation day (0 d 5.1 ± 1.03; P<0.0001). Results indicate there is variation in temperature and micro-behaviors in the days leading up to ovulation as well as between CON and OPEN cows.