Lab-Made Electronic Nose for Fast Detection of Listeria monocytogenes and Bacillus cereus

Abstract

The aim of this study is to determine the performance of a lab-made electronic nose (e-nose) composed of an array of metal oxide semiconductor (MOS) gas sensors in the detection and differentiation of Listeria monocytogenes (L. monocytogenes) and Bacillus cereus (B. cereus) incubated in trypticsoy broth (TSB) media. Conventionally, the detection of L. monocytogenes and B. cereus is often performed by enzyme link immunosorbent assay (ELISA) and polymerase chain reaction (PCR). These techniques require trained operators and expert, expensive reagents and specific containment. In this study, three types of samples, namely, TSB media, L. monocytogenes (serotype 4b American Type Culture Collection (ATCC) 13792), and B. cereus (ATCC) 10876, were used for this experiment. Prior to measurement using the e-nose, each bacterium was inoculated in TSB at 1 × 10³-10⁴ CFU/mL, followed by incubation for 48 h. To evaluate the performance of the e-nose, the measured data were then analyzed with chemometric models, namely linear and quadratic discriminant analysis (LDA and QDA), and support vector machine (SVM). As a result, the e-nose coupled with SVM showeda high accuracy of 98% in discriminating between TSB media and L. monocytogenes, and between TSB media and B. cereus. It could be concluded that the lab-made e-nose is able to detect rapidly the presence of bacteria L. monocytogenes and B. cereus on TSB media. For the future, it could be used to identify the presence of L. monocytogenes or B. cereus contamination in the routine and fast assessment of food products in animal quarantine.

Keywords: Bacillus cereus; LDA; Listeria monocytogenes; QDA; SVM; electronic nose.

Figure 1

(A) E-nose measurement set-up. I: a personal computer with a software of DAQ and chemometric models, II: sample on the hot plate, and III: the main part of e-nose; (B) III: main part of electronic nose device. a: sampling system, b: DAQ and controller, c: sensor chamber.

Figure 2

(a) Electrical schematic of recording a signal from a sensor in this e-nose. V_H and V_S are voltage sources for heater and sensor, respectively; RS and R_L are sensor resistance and load resistance, respectively; R and C for the low-pass filter. (b) Typical of gas sensor response of e-nose during the delay, sampling and purging processes.

Figure 3

(a) Listeria monocytogenes on listeria selective agar (LSA) media (b) Christie Atkins Munch Peterson (CAMP) test result; Bacillus cereus colonies (c) on mannitol egg yolk polymixin agar (MYP) media (d) on BCA media.

Figure 4

Radar plot of the average sensor responses obtained with the gas sensor array for each bacterial sample. N: TSB uninoculated; L: TSB inoculated with L. monocytogenes, B: TSB inoculated with B. cereus.

Figure 5

Classification of group N, B, and L after incubation for 48 h using linear discriminant analysis1 (LDA1). (a): N vs. B and (b): N vs. L, and (c) B vs. L.

Figure 6

Confusion matrix of two groups with different chemometric models of LDA (a–c), quadratic discriminant analysis (QDA) (d–f), and support vector machine (SVM) (g–i). The number in parenthesis indicates the accuracy.