New rapid viral plaque detection system, aided by deep learning and holographic imaging, can help accelerate vaccine and drug development 


Figure: Viral plaque detection system enabled by holography and deep learning. (Image courtesy: Aydogan Ozcan/UCLA)

In a new paper published in Nature Biomedical Engineering, a team of scientists led by Professor Aydogan Ozcan from the Electrical and Computer Engineering Department at UCLA and an associate director of the California NanoSystems Institute, developed a rapid, stain-free, and automated viral plaque detection system enabled by holography and deep learning. This system incorporates a cost-effective and high-throughput holographic imaging device that continuously monitors the unstained virus-infected cells during their incubation process. At each imaging cycle, these time-lapse holograms captured by the device are periodically analyzed by an AI-powered algorithm to automatically detect and count the viral plaques that appear due to virus replication.

The proof-of-concept and effectiveness of this system were demonstrated using three different types of viruses: vesicular stomatitis virus (VSV), herpes simplex virus type-1 (HSV-1), and encephalomyocarditis virus (EMCV). By utilizing this system, UCLA researchers achieved the detection of over 90% of VSV viral plaques within 20 hours of incubation without any chemical staining, demonstrating a time saving of more than 24 hours in comparison to the traditional plaque assay, which requires 48 hours of sample incubation. In the case of HSV-1 and EMCV, this system effectively reduced their viral plaque detection times by approximately 48 and 20 hours, respectively, compared to the detection time needed for the traditional staining-based viral plaque assay.

In addition to offering major time savings, this stain-free and cost-effective system can successfully identify individual viral plaques within clusters as opposed to the traditional viral plaque assays, which fail to separately detect and count those individual plaques within clusters due to the spatial overlap of their signatures.

Tairan Liu, Yuzhu Li, Hatice Ceylan Koydemir, Yijie Zhang, Ethan Yang, Merve Eryilmaz, Hongda Wang, Jingxi Li, Bijie Bai, Guangdong Ma, and Aydogan Ozcan

Aydogan Ozcan: “Our results and analyses highlight the transformative potential of this AI-powered viral plaque detection system to be used with various plaque assays in virology, which might help to expedite vaccine and drug development research by significantly reducing the detection time needed for traditional viral plaque assays and eliminating chemical staining and manual counting entirely.”

This study is published online in the journal Nature Biomedical Engineering.

National Science Foundation