A close to real-time prediction method of total coliform bacteria in foods based on image identification technology and artificial neural network

Abstract

A prediction method of total coliform bacteria based on image identification technology in foods was proposed. In order to get the close to real-time detection results, this method used the total count of bacteria and bacilli to predict the total coliform bacteria counts because coliforms are difficult to extract the feature parameters to be recognized and enumerated, while total count of bacteria and bacilli could be enumerated by using image identification technology. An optimal artificial neural network (ANN) model was presented for prediction of total coliform bacteria counts. Several configurations were evaluated while developing the optimal ANN model. The optimal ANN model consisted two hidden layers with five neurons in each hidden layer. Results showed that predicted total coliform bacteria counts were positively correlated to the experimental total coliform bacteria counts obtained by traditional multiple-tube fermentation technique (correlation coefficient, R2 = 0.9716), which predicted accuracy was much better than other predicted models (the correlation coefficient of linear regression model, second-order polynomial regression model and polynomial trend surface analysis was 39.81%, 67.17% and 78.85%, respectively).