Development of an automated E. coli colony counting system for faster enumeration

Abstract

Manual counting of E. coli colonies, a critical process in microbial water quality assessment, is hindered by significant limitations, including high time consumption, human error, fatigue, and variability in visual acuity. These challenges often lead to inaccurate colony counts, potentially compromising public health outcomes in water and sanitation monitoring. This study addresses these issues by developing an automated system that can count and identify E. coli colonies. This system integrates computer vision algorithms that can detect and count E. coli colonies by a mechanism of change of light intensity across the filter paper. 204 images of filter paper that were subjected to the membrane filtration method in a Mukono water point sampling done by Aquaya in June 2024 were subjected to this system. Computer vision algorithms were developed using Python, leveraging OpenCV libraries for image processing tasks such as colony detection and segmentation. The machine learning model, implemented in Visual Studio Code, was trained to identify and count colonies based on pixel intensity and morphological features. A user-friendly interface was created using the React framework to facilitate interaction with the system, enabling seamless data input and result visualization. The system was tested on a dataset of filter paper images with varying E. coli colony densities, and performance was evaluated against manual counts conducted by Aquaya research assistants. A t-test was carried out and p value of 0.83 obtained signifying strong positive correlation and a correlation coefficient of 0.76. The developed automated E. coli colony counting system to a great extent addresses the limitations of manual methods by offering a faster, more accurate, and less labor-intensive solution. By integrating computer vision and a user-friendly interface, the system reduces human error and fatigue and rate of processing. Its accuracy and processing time make it a valuable tool for organizations in the water and sanitation sectors, enhancing microbial monitoring capabilities and supporting improved public health outcomes. Future work will focus on optimizing the system for real-world field conditions and expanding its applicability to other microbial species.