Odour-based heat detection in dairy cows: a pilot laboratory study

Salla Ruuska, Antti Roine, Heli Lindeberg, Pekka Kumpulainen, Mikko Järvinen, Niku Oksala, Jouko Vepsäläinen6, Jaakko Mononen

    Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review


    Current automated heat detection methods for dairy cattle are based on activity meters or milk progesterone analyses. The profile of volatile pheromone compounds varies during the oestrous cycle. An eNose is a device developed for analysis of gas mixtures and could potentially be used to detect heat, as well as health problems such as sub-clinical ketosis, in cattle. In the present pilot laboratory study, we investigated whether eNose could discriminate odour samples (vaginal fluid) between oestrous and non-oestrous dairy cows. The odour samples (OS) were collected from 13 dairy cows by rotating a cotton tipped applicator in the vagina (minimum depth of 7.5 cm) through a complete revolution in each direction before withdrawing. During the odour sampling, the cows were either in dioestrus (between Day -14 and Day -8, 6 OSs), in oestrus (Day 0, 28 OSs), or pregnant (Day 42, 26 OSs). Oestrus was defined as the day of standing heat (a cow allowed other herd mates to mount her while she remained standing). Dioestrus and pregnancy were defined by rectal palpation of ovaries and uterus confirmed with ultrasound examination. The OSs were kept in a freezer (−20 °C) and thawed on the day of the laboratory analyses. Each OS was analyzed using ChemPro 100i –eNose (Environics Inc., Mikkeli, Finland) which yields a smell print with 18 variables. A Sammon projection (Figure 1) shows that the smell prints of the cows in oestrus are in one cluster and the smell prints of the cows in dioestrus or pregnant in another cluster. Logistic regression was used to identify a classifier that discriminates the former cluster from the latter cluster. Leave-one-out-cross-validation was used to validate the results. After cross-validation, the classifier correctly classified 90% of the samples, resulting in a sensitivity of 86% and specificity of 93%. The results show that eNose has considerable potential for heat detection. However, further biological validation with a larger data set is needed. Future studies should focus also on development of less invasive sampling such as analysis of milk headspace or development of environmental sensor that simply sniffs the animal.
    Original languageEnglish
    Title of host publicationProceedings of the Second DairyCare Conference 2015
    Number of pages1
    Publication statusPublished - 2015
    Publication typeA4 Article in conference proceedings
    EventDairyCare Conference - Cordoba, Spain
    Duration: 3 Mar 20154 Mar 2015


    ConferenceDairyCare Conference


    Dive into the research topics of 'Odour-based heat detection in dairy cows: a pilot laboratory study'. Together they form a unique fingerprint.

    Cite this