Meat (beef) quality and safety evaluation using electronic nose systems

Two types of electronic nose systems were evaluated for their performances to identify spoiled and contaminated meat samples. One of the electronic nose systems (P-module E-nose system, Cyranose-320(TM)) contained an array of 32 conducting polymer sensors. The other nose system (M-module E-nose system) had an array of 9 metal oxide sensors. The meat samples (M. Longissimus lumborum) were packaged simulating the existing retail grocery package conditions and were stored at two different temperatures (10°C and 4°C).; Once the signals were acquired by the electronic nose systems from the headspace of the meat packets, they were pre-processed to reduce the noise, and features were extracted. Linear (LDA) and quadratic discriminant analysis (QDA) based classification models were developed from the extracted features. These models were validated by two techniques: leave-1-out method and bootstrapping.; For the meat spoilage identification studies, the developed models classified meat samples based on the microbial population: "unspoiled" (microbial counts <6.0 log10 cfu/g) and "spoiled" (microbial counts >6.0 log 10 cfu/g). For the meat contamination studies, the developed models classified meat samples into two classes: "No Salmonella" (microbial counts < 0.7 log10 cfu/g) and "Salmonella " (microbial counts > 0.7 log10 cfu/g).; The spoilage experiments yielded maximum classification accuracies of approximately, 97.0% and 98.0%, respectively, for the meat samples stored at 10°C and 4°C. Both the P-module and M-module E-nose systems yielded similar maximum total classification accuracies.; When the electronic nose systems were used for identifying Salmonella inoculated beef, the maximum total classification accuracy obtained was between 80.5% and 87.3% for the meat samples stored at 10°C and 4°C, by using both the E-nose systems independently. As in the case of the spoilage experiments, QDA and bootstrapping method of data analysis provided the maximum classification accuracies. The experimental results obtained so far show promising trends for the implementation of electronic nose systems as intelligent sensing systems for identifying meat spoilage and contamination. Further research in this direction is recommended by validating the results on larger data sets, including additional features and utilizing other higher order data analysis techniques for building the classification models.