Timely and accurate determination of the presence and nature of foodborne bacterial pathogens is of utmost importance in food safety. In this paper, we investigated whether an electronic nose (based on a metal-oxide sensor array) could distinguish between E. coli and Listeria. Unlike other studies in this area, samples consisted of individual colonies transferred from agar plates, then suspended in phosphate buffered saline. Features extracted from the sensor response curves capture both static (steady state) and dynamic (slope) information. The use of a linear classifier in association with supervised dimensionality reduction, using uncorrelated linear discriminant analysis (ULDA), yielded classification accuracies of 92.4%. The proposed method has the potential to reduce the overall time required to identify bacterial pathogens. This type of sample is available relatively early in the inspection process, so discrimination based on the odour signature of single colonies has the potential to reduce time and cost by eliminating or reducing subsequent culturing stages and biochemical testing. The results presented herein suggest that further research in this area is warranted, particularly with a wider variety of bacterial species.

Additional Metadata
Keywords Bacteria identification, Classification, Dimensionality reduction, Electronic nose, Food inspection
Persistent URL dx.doi.org/10.1016/j.snb.2010.09.062
Journal Sensors and Actuators, B: Chemical
Citation
Green, G.C. (Geoffrey C.), Chan, A, Dan, H. (Hanhong), & Lin, M. (Min). (2011). Using a metal oxide sensor (MOS)-based electronic nose for discrimination of bacteria based on individual colonies in suspension. Sensors and Actuators, B: Chemical, 152(1), 21–28. doi:10.1016/j.snb.2010.09.062