Summary:
UCLA Researchers in the Department of Electrical and Computer Engineering have developed a thin-film transistor-based image sensor that can quantify and identify bacterial colony forming units (CFUs) with high accuracy using a deep-learning algorithm.
Background:
Bacterial infections are a leading cause of death every year in both developed and developing nations, accounting for millions of deaths annually. The associated expense of treating bacterial infections cost more than $4 billion annually in the United States alone. Among bacterial infections, Escherichia coli (E. coli) and other coliform bacteria are the most common ones and typically transmitted via fecal contamination in water. The basic and most frequently used EPA-approved methods to test for the presence of these bacteria (EPA 1103.1 and EPA 1604) are bacterial-culture based methods. These traditional techniques take at least 24 hours for the read-out and necessitate expertise from the trained microbiology professionals for the final results. Nucleic acid-based approaches have been developed that produce results in less than a few hours, but these methods suffer from poor sensitivity and difficulty differentiating live versus dead bacteria—and thus are not EPA approved for screening water samples. Other rapid identification technologies exist, but they suffer from limits in the volume of water that can be tested at once (< 0.1 L) or are operationally intensive to use. There is a growing need for a rapid bacteria colony determination method that is both operationally simple and capable of high throughput detections.
Innovation:
UCLA researchers have recently developed a thin-film transistor (TFT) based imaging sensor in conjunction with deep learning algorithms to rapidly identify and automatically count bacterial colonies and species on agar plates. The sensing system has an ultra-large imaging field of view and does not require mechanical scanning making it cost-effective, operationally simple to use, and portable. In addition to the foregoing advantages, this TFT based imaging method is free of any spurious result from the image processing step. The combination of TFT-based imaging and the state-of-the-art machine learning techniques enable the differentiation between dead and living bacteria and can properly identify bacterial colonies in as little as six hours with high accuracy—minimizing identification time by up to 12 hours. Additionally, this technology can easily be integrated into agar plates, opening new opportunities for microbiology laboratory instrumentation.
Potential Applications:
• Rapid bacteria identification
• Portable biosensing
• Antimicrobial susceptibility testing
• Water safety testing
Advantages:
• Cost-effective
• Operationally simple
• High throughput
• Lens-free imaging
• Large Field of view
• Rapid and automated detection
• Algorithm-based identification
Development to Date:
Successful demonstration of the invention demonstrated.
Related Papers: Li, Y.; Liu, T.; Koydemir, H. C.; Wang, H.; O’Riordan, K.; Bai, B.; Haga, Y.; Kobashi, J.; Tanaka, H.; Tamaru, T.; Yamaguchi, K.; Ozcan, A. Deep Learning-Enabled Detection and Classification of Bacterial Colonies Using a Thin-Film Transistor (TFT) Image Sensor. ACS Photonics 2022, 9 (7), 2455–2466. https://doi.org/10.1021/acsphotonics.2c00572.
Reference: UCLA Case No. 2022-258
Lead Inventor: Aydogan Ozcan, Yuzhu Li, Tairan Liu