Waterborne diseases affect more than 2 billion people worldwide, which causes economic difficulties: the treatment of all these diseases requires resources. For example, in the United States, waterborne diseases cost more than $2 billion a year, and 90 million cases are reported each year. Traditional methods for detecting bacteria in water take too much time and scientists decided to rectify the situation. Details about their development are published by the magazine Light: Science & Applications.
Among water-related pathogens, one of the most common public health problems is the presence of E. coli bacteria in drinking water.
Traditional culture-based bacteria detection methods often take 24-48 hours, followed by visual inspection and colony counting by an expert in accordance with the US Environmental Protection Agency (EPA) guidelines.
Alternatively, molecular detection methods based, for example, on the amplification of nucleic acids, can shorten the detection time to several hours. However, they usually lack sensitivity to detect bacteria at very low concentrations. Also, they are not able to distinguish between living and dead microorganisms. In addition, there is no EPA-approved nucleic acid method for detecting coliform bacteria in water samples.
Therefore, there is an urgent need for an automated method with high sensitivity, which can provide fast and high-performance detection of bacterial colonies. This will provide an alternative to the methods currently available by the EPA that take at least 24 hours and require an expert on colony counting.
As a result, a group of scientists led by Professor Aydogan Ozkan of the Department of Electrical and Computer Engineering at the University of California, Los Angeles (UCLA), USA, and their colleagues developed an intelligent AI-based imaging system for the early detection and classification of living bacteria in water samples.
Based on holography, the researchers developed a highly sensitive and high-performance imaging system. It captures microscopic images of the entire bacterial culture bowl. This is necessary to quickly detect colony growth by analyzing slow-motion images of the neural network. After detecting the growth of each colony, a second neural network is used to classify the type of bacteria.
The effectiveness of this unique platform has been demonstrated by the early detection and classification of three types of bacteria, namely E. coli, Klebsiella aerogenes, and Klebsiella pneumoniae. UCLA researchers have reached the limit of detection of 1 colony-forming bacteria per 1 liter of water sample for 9 hours of total testing time. In doing so, they demonstrated time savings for detecting bacteria over 12 hours compared to the standard EPA method.
These results highlight the potential of the new AI-based holographic imaging platform, which not only provides highly sensitive, fast, and cost-effective detection of live bacteria but also provides a powerful and versatile tool for research in microbiology.