Visible and near infrared reflectance spectroscopy (NIRS) for the assessment of flesh foods
Visible and near infrared (NIR) reflectance and transmission spectra (400 - 2500 nm) of various flesh foods in various presentations were examined for qualitative and quantitative analysis. Discriminant functions for muscle types and animal species were included. Lamb muscles (n: 306), chicken breast and thigh muscles (n: 48), bull and steer muscles (n: 103), raw fish (n: 80), fish meal (n: 700) and fish oil (n: 160) samples were examined in the experiments. INTACT and MINCED presentation to the instrument were compared, as well as type of muscle (longissimus dorsi, infra spinatus, supra spinatus, semimembranosus, semitendinosus, rectus femoris), effect of sex and both longitudinal (LS) and transverse section (TS) of the muscle on the optical properties and Partial Least Squares (PLS) calibrations for gross composition. MINCED presentation to the instrument gave the best results for the NIRS calibrations in all the muscle tissues utilized, while, INTACT presentation showed poorer calibrations. Muscle type and sex affect the calibrations. Raw fish and fish by-products, both fish meal and oil had good NIRS calibrations for the twelve parameters analyzed. The results show that NIRS is acceptable as a method for determining gross composition in a wide variety of flesh foods in MINCED presentation. Principal component analyses (PCA) and Soft Independent Modelling of Class Analogy (SIMCA) were used for the authentication and classification between muscles and among animal species. The conclusions of this work are that NIRS can successfully predict chemical composition in different muscles in MINCED rather than INTACT presentation. Classifications using PCA and SIMCA were excellent tools to authenticate flesh foods.