Stochastic resonance-based electronic nose: A novel way to classify bacteria

An electronic nose (e-nose), the Cyrano Sciences’ Cyranose 320 comprising an array of 32 polymer carbon black composite sensors, has been used to identify six species of bacteria responsible for eye infections when present at a range of concentrations in saline solutions. Readings were taken from the headspace of the samples by manually introducing the portable e-nose system into a sterile glass containing a fixed volume of bacteria in suspension. All bacteria were grown on blood or lysed blood agar in standard Petri dishes at 37 °C in a humidified atmosphere of 5% CO2 in air. After overnight culturing, the bacteria were suspended in sterile saline solution (0.15 M NaCl) to a concentration of approximately 108 colony forming units (cfu)/ml. A 10-fold dilution series of bacteria in saline was prepared and three dilutions (d1 = 108 cfu/ml, d2 = 105 cfu/ml and d3 = 104 cfu/ml) were sniffed using the e-nose. Gathered data were a very complex mixture of different chemical compounds. After some pre-processing, gathered data for different bacteria were passed through individual single threshold signal detectors with added noise. Randomly added Gaussian noise (the noise intensity which is the ratio of the standard deviation of noise to the standard deviation of signal varied from 0 to 4) was in the same range of magnitude for all classes of bacteria. In this single threshold-based system stochastic resonance (SR) occurred; which resulted in an enhancement, by noise, of the response of the e-nose system to the gathered bacterial signals. For different bacteria classes “maximum cross-correlation coefficients” were found to be completely different; this ‘maximum cross-correlation coefficient’ was found to be constant for a particular class of bacteria. It was evident that these ‘maximum cross-correlation coefficient’ can be used to accurately represent the different classes of bacteria. Then three supervised classifiers, namely multi-layer perceptron (MLP), probabilistic neural network (PNN) and radial basis function network (RBF), were used to classify the six bacteria classes. A comparative evaluation of the classifiers was conducted for this application. The best results suggest that we are able to predict the six classes of bacteria with up to 98% accuracy with the application of stochastic resonance along with RBF network. We believe that these results show that there is good potential for the use of SR to improve system performance in similar applications.