Escherichia Coli Bacterial Identification using Deep Learning Approach

Urinary Tract Infections (UTIs), predominantly caused by Uro-Pathogenic Escherichia Coli (UPEC), account for 80% of cases, posing a significant health challenge. Traditional methods for bacterial identification, such as culturing, are both time-consuming and costly. In this study, a Deep Learning (DL) approach using a Convolutional Neural Network (CNN) model developed with TensorFlow’s Keras API is presented to quickly and accurately identify E. coli and non-E. coli bacteria colonies from a custom dataset of 978 images collected using mobile devices. The model, comprising three convolutional layers (32, 64, and 128 filters), was trained for 30 epochs, achieving 91% accuracy. Max pooling, dropout layers, $\mathbf{L} 2$ regularization and early stopping was implemented to prevent overfitting. Data augmentation was also applied to enhance model generalization. The findings demonstrate that this cost-effective, real-time deep learning method provides an efficient alternative to conventional identification techniques, offering a promising tool for rapid diagnostics in clinical environments.