Researchers apply AI to medical imaging to learn about COVID-19
When COVID-19 began spreading across the world, scientists and doctors had to analyze a flood of new information about the unknown virus. The rapid influx of new data about the disease provided the perfect dataset to one community of researchers: medical imaging specialists interested in the growing field of artificial intelligence.
In August 2020, the National Institute of Biomedical Imaging and Bioengineering announced a new AI initiative, the Medical Imaging and Data Resource Center. The center enables researchers to apply AI to medical imaging, which uses technology to create images of the interior of a body. This work has led the center to create a collaborative community that seeks to improve medical AI algorithms while furthering research on diseases like COVID-19.
A rapid response
The medical imaging community had already been exploring new ways to use AI, and collaborating through conferences, workshops and research articles years before the pandemic.
These activities laid the groundwork for researchers to rally around the creation of the center when the pandemic arrived. By August 2020, the biomedical imaging institute had awarded a two-year contract to launch the new center at the University of Chicago. Dr. Krishna Kandarpa, the institute’s director of research sciences and strategic directions, said that the contract for the center was reviewed thoroughly, but that “we did this much faster than would occur in normal times.”
How does the center integrate AI with medical imaging?
Researchers at the center collected an initial batch of about 10,000 medical images, stripped of personally identifiable information, from two radiological organizations that had already gathered medical images of COVID-19 patients. The organizations narrowed their focus to two organs primarily affected by COVID-19: the lungs and the heart.
The center curates, annotates and stores the images in a secure network that is accessible to other researchers working in the field. These researchers then use their own AI algorithms and tools to analyze the information.
“It is basically a sandbox where everybody can write their AI algorithms and test them on this large repository,” said Kandarpa.
How exactly does AI help streamline COVID-19 research?
Physicians use medical imaging to detect, diagnose and understand diseases. AI amplifies this process by analyzing thousands of pictures and data points much quicker than any individual scientist could. AI can also synthesize data from different medical centers and find patterns that could indicate the existence of new disease strains or variants.
By supporting new uses of AI, the center will contribute to the fight against COVID-19, helping researchers quickly and efficiently answer several important questions about the virus: What is the likely course of symptoms for different groups of patients? Are there signs that point to the disease recurring in a certain type of patient? And how do different treatments affect the disease itself?
But the center’s work to combat COVID-19 is only the beginning. The Food and Drug Administration helped create the center, meaning that some of the AI algorithms researchers employ to analyze the center’s medical images could one day become integral to regulated, approved and widely used medical devices. Widespread use of AI in the field could transform medical imaging, help answer questions about other existing diseases and help researchers swiftly respond to any future medical crises.
For more details about the Medical Imaging and Data Resource Center—along with other stories of how government used technology to navigate the COVID-19 pandemic—read the Partnership’s research report, Bit by Bit: How governments used technology to move the mission forward during COVID-19.
Will Butler is a former intern on the Partnership’s Communications team.