Classification par spectroscopie proche infrarouge des gros bovins ou des carcasses suivant le profil en acides gras des viandes
Résumé
The aim of the study was to predict by NIRS (near infrared spectroscopy) the compliance of cattle or carcass to a
specification based on beef fatty acids (FA) without predicting individual FA. The leathers of 380 cattle from various
genetic types (milk or meat), categories (young bulls, heifers, steers or cows) and rearing systems were scanned by
NIRS portable equipment. After slaughter, Rectus abdominis was scanned on each carcass and sampled. FA
compositions were assessed using gas chromatography and spectra were measured at wavelengths between 350 and
2500 nm. The performances of prediction models are satisfactory with a sensitivity of 78%, a specificity of 80% and an
accuracy of 79% from carcass spectral measurements, and respectively of 85%, 74% and 80% from animal spectral
measurement.