Development and evaluation of gustatory and olfactory sensors for detection of Salmonella contamination in packaged beef

Detection of pathogen contamination inside the food matrix is a major challenge. In this regard, a sensor fusion approach was applied to develop and evaluate the performance of olfactory and gustatory sensors for detection of pathogens inside packaged beef. A commercial electronic nose system was evaluated for its ability to detect indicator compounds associated with meat safety. This system was used to identify selective sensors sensitive to specific compounds at various concentrations. Different vapor concentrations of ethanol ranging between low to high detection limits were tested to evaluate the performance of the commercial system. It was observed that a few sensors (4, 12, 15, and 28) showed a significant difference between 38 ppm and 300 ppm for acetic acid. Linear Discriminant Analysis (LDA), along with bootstrap, provided the highest classification accuracies of 88.33% for 100 ppm and 200 ppm of acetic acid. However, Quadratic Discriminant Analysis (QDA) provided a higher classification accuracy of 83.64% at lower concentration levels of 38 ppm and 75 ppm. Similarly, QDA was a better model of choice for lower concentrations of ethanol (37 ppm and 100 ppm) providing a higher classification accuracy of 86.83%.;Another approach of the current research involved the development and evaluation of gustatory sensors for detection of Salmonella contamination in packaged beef. Performance of different Surface Enhanced Raman spectroscopy (SERS) substrates was evaluated to discriminate between the control and spiked samples. Preliminary studies were conducted using commercial silver doped sol-gel vials and silver sol prepared in the laboratory. The commercial Real Time Analyzer (RTA) vial provided an average classification accuracy of 90.25% between control and spoiled samples. A laboratory prepared SERS substrate was also used for the identification of control and spiked beef samples. This technique provided a classification accuracy of approximately 80.35% between two groups.;Labeled and label free techniques were used to evaluate the performance of commercial off-the-shelf (Klarite) SERS substrate. It was found that the labeled technique provided better distinction between control and spiked groups than the label free technique. The SERS effect using silver sol showed potential to detect acetone to a LDL of 9.78 ppm.