Application of udder surface temperature by infrared thermography for diagnosis of subclinical mastitis in Holstein cows located in tropical highlands

Several reports have indicated that udder surface temperature (UST) can be a useful indicator of subclinical mastitis (SCM). The objective was to evaluate UST by infrared thermography (IRT) as a diagnostic tool for SCM and intramammary infection (IMI), and to assess the influence of environmental conditions in the potential diagnosis of this disease in dairy cows located at high-altitude tropical regions. A total of 105 cows (397 quarters) from 3 dairy farms with mechanical and manual milking methods were enrolled in the study. Subclinical mastitis was diagnosed when quarter samples had a somatic cell count (SCC) ≥200 × 10³ cells/mL, microbial growth (MG) was defined when a major pathogen (≥1 cfu/plate) or Corynebacterium spp. (≥10 cfu/plate) was isolated, and IMI was defined as the presence of MG and SCC ≥100 × 10³ cells/mL. Infrared images were taken with a thermal camera placed 1 m away from the udder, and shots of the rear and left and right lateral view were made during the morning milking, before any manipulation of the udder and employing dark cardboard on the contralateral side to avoid artifacts in the background. A multilevel mixed effects linear regression model clustered within cows and herd was performed to evaluate the associations with UST. Clinical performance was evaluated using the Youden index to establish the optimum UST thresholds, which were set at 32.6°C for any case definition when milking was by hand, at 33.7°C for MG, and at 34°C for SCM and IMI in machine-milked quarters. Sensitivity (Se), specificity (Sp), area under curve (AUC), and positive likelihood ratio (+LR) were also assessed. Test agreement was assessed by kappa coefficient (κ). The UST of healthy quarters ranged between (95% CI) 32.4 and 32.6°C, lower than SCM quarters (n = 88) at 32.9°C (95% CI: 32.7-33.1 °C), MG quarters (n = 56) at 33.5°C (95% CI: 33.3-33.7°C), and IMI quarters (n = 50) at 33.5°C (95% CI: 33.2-33.7 °C). The UST was also related to the milking method: higher temperatures were observed for hand milking (n = 90) compared with machine milking (n = 185). No relation between environmental conditions such as wind speed, atmospheric temperature, relative humidity, and temperature-humidity index and UST were observed during this study. For hand milking, the optimal UST threshold was 32.6°C; for SCM, Se = 0.53, Sp = 0.89, AUC = 0.71, κ = 0.4; for MG, Se = 0.83, Sp = 0.93, AUC = 0.88, κ = 0.77; and for IMI, Se = 0.82, Sp = 0.92, AUC = 0.87, κ = 0.74. The machine milking threshold for SCM resulted in Se = 0.42, Sp = 0.97, AUC = 0.70, κ = 0.47; for MG, Se = 0.82, Sp = 0.89, AUC = 0.85, κ = 0.60; and for IMI, Se = 0.82, Sp = 0.98, AUC = 0.90, κ = 0.79. These findings suggest that UST determined by IRT is higher in machine-milked cows and in quarters with MG and IMI than in healthy quarters; therefore, UST by IRT is a reliable, clinically useful method for MG and IMI diagnosis.