Purpose: The purpose of this study was to investigate the potential of NIR (Near Infrared) spectroscopy, as a rapid, non-destructive and sufficiently accurate method for detection and quantification of microbial load in goat milk.
Methods: For that, collected milk samples were incubated at different temperatures (5, 10 and 15°C) and microbial concentration changes were measured over incubation time by plate count method (PCM) and NIR technique based on diffuse reflectance integrating sphere in a region 1100 nm – 2500 nm. The thickness of the sample was 0.5 mm. Different models were compared on 50 samples in range 3.7-8.6 log CFU/ml. Final predictive models were developed based on continuum regression, and parameters were set in a two-stage optimization procedure.
Results: Results indicated a good performance of the model to predict bacterial contamination in milk over time as shown by the Mean Square Error computed on independent validation set, whose value corresponded to 0.5 log CFU/ml. This is an encouraging result, considering variability of verification method (i.e., PCM).
Significance: Results in this work suggest NIR technology as a suitable method for rapid detection of bacterial contamination, improving efficiency of the production process and food safety of this commodity.