Identification of breast defects in broilers with short-wave pocket near-infrared spectrometer

Chicken breast meat is the most popular animal protein source, but three novel myopathies have been described in broilers over the last decade known as wooden breast (WB), white striping (WS) and spaghetti breast (SB). This study assessed the potential of a pocket-sized near-infrared (NIR) spectrometer to predict WB, WS and SB in broilers. A total of 4313 breasts from Ross 308 strain chicks were scanned with the pocket-sized NIR device from 740 to 1070 nm. Each spectrum was matched with the defects information, and a partial least square discriminant analysis was conducted with external validation. Results revealed that breast defects reduced the sample’s absor bance. Neither the spectrum first derivative nor wavelength selection improved the models’ accuracy compared to the raw spectrum. The highest balanced accuracy was obtained for detecting WS (0.80), followed by WB (0.75) and SB (0.51). All models achieved remarkably high sensitivity (>0.89), indicating their proficiency at identifying samples without defects. On the other hand, the specificity was low (<0.64). In conclusion, models developed using raw absorbance can be used without any additional mathematical treatments to identify samples without defects. Further studies are suggested to confirm if positive samples are genuinely affected by a myopathy. Funded EU H2020 101000250-INTAQT