Rapid prediction of the chemical composition of pet food using a benchtop and handheld near-infrared spectrometer

Abstract

Quality of pet foods can be affected by many factors such as raw materials, formulations, and processing techniques. The pet food manufacturers require fast analyses to control the nutritional quality of their products. Herein, near-infrared spectroscopy (NIR) was evaluated to quantify the chemical composition of pet food, and the performances of two NIR spectrometers were investigated and compared: a benchtop instrument (1000–2500 nm) and a low-cost handheld instrument (900–1700 nm). Seventy cat food and thirty-six dog samples were characterized using reference methods for crude protein, crude fat, crude fibre, crude ash, moisture, calcium (Ca), and phosphorus (P). Principal component regression (PCR) and partial least squares regression (PLSR) were used to establish the models that involved the cat food and mixed model. The characteristic wavelengths were selected using a competitive adaptive reweighted-sampling (CARS) algorithm. The Optimal models obtained from the benchtop instrument for crude protein, crude fat, and moisture were classified as “Good” or “Very good” (Residual prediction variation (RPD) > 3), for crude fibre were classified as “Poor” (RPD>2), and for crude ash, Ca and P (RPD<2) were classified as “Very poor”. The Optimal calibrations obtained from the handheld instrument for crude protein, crude fat, and moisture were classified as “Good” or “Very good” (RPD>3), for crude fibre, crude ash, Ca, and P were classified as “Very poor” (RPD<2). Generally, the the performance of benchtop and handheld instrument was close, and the cat food model outperformed the mixed model. Results from the current study revealed the potential to monitor the chemical compositions in pet food on a large scale.