M. Boukadoum (Canada), A. Bensaoula, and D. Starikov (USA)
neural network, bacteria detection, fluorescence measurement, light emitting diodes.
We describe a device that can serve to identify the presence and concentration of live bacteria in a liquid environment. The method is based on fluorescence measurements from fluorophore bacteria, or bacteria tagged by fluorescent protein markers, and by processing the acquired data with a neural network for spectral signature identification and concentration measurement. In this work, we studied tagged E-coli and our neural network consisted of a multilayer perceptron (MLP) with one hidden layer and one output neuron, trained with the Resilient Backpropagation (RPROP) algorithm. Our results show that this system is efficient at distinguishing bacteria from several non biological, organic luminescent substances, and that it can also serve for the quantitative evaluation of the bacteria concentrations.
Important Links:
Go Back
