| dc.description.abstract | Milk is a food with significant nutritional value and for being rich in nutritional constituents is presented as a great growth medium for microorganisms. Therefore, the "raw" milk, must undergo a thermal treatment to reduce or eliminate the most harmful bacteria. The consumption of milk that has passed through improper treatment, can lead to infections due to various pathogens such as, Salmonella spp.. For being highly detrimental, the presence of these bacteria in food is intolerable, making the microbiological food testing’sto determine the presence of this microorganism, absolutely required. Conventional methods to detect Salmonella spp. are laborious, requiring the use of significant amounts of liquid and solid media and reagents, besides demanding time-consuming procedures. Therefore, alternative methods which provide advantages, such as, relatively low cost, fast, non-destructive, free of reagents and, consequently, which does not generate waste are urgent demand did to the quality control of this kind of food. Thus, this research aimed to offer an alternative methodology based on Near Infrared (NIR) spectroscopy to discriminate between the presence and absence of Salmonella spp. in whole and skimmed UHT milk. Different samples of whole and skimmed milk, both treated through UHT were submitted to data analysis conducted in Matlab ® software, using PLS toolbox® tools. The PLS-DA models were median centered and cross-validated using leave one out algorithm. The model to discriminate skimmed milk samples was constructed using four latent variables and presented RMSEC = 0.1639; RMSECV = 0.2084 and RMSEP = 0.0971. To the whole milk, the model was built using six latent variables with RMSEC = 0.1351 values; RMSECV = 0.2076 and RMSEP = 0.0928. The results showed that the suggested methodology was able to differentiate between contaminated and uncontaminated samples successfully, presenting potential to be implemented in the quality control sector of milk industry. | |
