Infra-red thermography as a monitoring tool for detection of sub clinical mastitis in dairy ewes

Infra-red thermography (IRT) is a no contact measurement technique of superficial temperatures (ST). It is used as a diagnostic or screening tool. In this study, sterile milk samples were taken from 236 clinically healthy milking ewes, one sample for each half-udder. They were analysed both for SCC as well as for bacteriological positivity. Sub-clinical mastitis was defined as: >500,000 SCC, and bacteriological test (+). Based on these results, animals with doubtful or incomplete data were ruled out, and the rest were divided into 3 groups: A: both half-udders (-),102 ewes. B: one half-udder (-) one half-udder (+), 50 ewes. C: Both half-udders (+), 30 ewes. IRT evaluation was done before sampling using a FLIR IR- vet, 420 esc. camera. The evaluation included 4 measurement areas; right and left half-udders with lower and upper area for each. Irrespective of the individuals, positive half-udders showed a significant difference of ST in the lower area compared to negative ones (P<0.02). However, it was not possible to define a temperature clearly indicative of a suspected subclinical mastitis. Nevertheless, comparing the delta temperature expressed in absolute value between the half udders in the same ewe showed a highly significant difference (P<0.0001) between the 3 groups. The ewes were then divided into 2 groups by placing a delta temperature cut-off (0.5 °C) in the lower mammary area. The results were as follows; 110 ewes resulted as ‘mastitis unsuspected’, 87 were correctly identified (based on the gold standard of bacteriological positivity) and 23 were false negatives. A further 72 ewes resulted as subclinical mastitis suspects (delta temp>0.5 °C), with 15 false positives and 57 correctly identified cases. Hence, adopting delta temperature as a criterion for animals suspected as having subclinical mastitis in at least one half udder showed a sensitivity equal to 71% (57/80), a specificity of 85.3% (87/102) and an accuracy of 79% ((57 + 87) / 182), while setting a cut-off at 0.5 °C.