Dr. Larry Zeng Places 3rd Worldwide in Low-Dose CT Contest
Raw low-dose image of the kidney compared to Dr. Zeng's improved version
Aug 31, 2016 12:00 AM
By Kirsten Mallik
The “Low-Dose CT Challenge” was an open global contest to solve one of medical imaging’s toughest problems. It started with 103 teams from 26 countries - and in the end, University of Utah Radiology and Imaging Sciences researcher Dr. Larry Zeng placed third in the world.
The challenge concerned CT (computer tomography) scanners: vital medical devices that help doctors see inside a patient by assembling multiple 2D cross-sectional images into a single 3D image. They are fast and efficient, needing just a few minutes to produce an image that shows bones, soft-tissue, and blood vessels all at the same time. CT scans are ideal for a radiologist looking for injuries, blood clots, or tumors.
However, CT scanners use x-ray radiation to create these images. And while each exposure is generally very low and very safe, repeat exposure over a lifetime can increase risk for some cancers, especially in women and children. However, right now, lowering the radiation dose means more visual distortion and “noise” in each image, making it harder for the radiologist to make a clear diagnosis.
This was the “Low-Dose CT Grand Challenge”, co-sponsored by the NIH, the Mayo Clinic’s CT Innovation Center, and the American Association of Physicists in Medicine (AAPM). Can we create a useful image with a lower dose of x-rays?
Dr. Larry Zeng is a mathematician and an engineer, and a faculty member of the Radiology and Imaging Sciences Department’s Utah Center for Advanced Imaging Research (UCAIR). He built an algorithm – like a mathematical photo filter - that uses geometry to rapidly detect and remove image “noise”, beautifully cleaning up distorted CT scans. Not only did his work receive the third highest score in the Global Challenge, but his algorithm could help enhance CT scans distorted for all kinds of reasons, not just lower radiation doses.
“Reducing radiation without sacrificing the speed or accuracy of a diagnosis is better for all patients – and Larry’s algorithm is a great step towards that goal.” Says Dr. Ed DiBella, director of UCAIR. “Improved CT images would help so many different branches of medicine.”
Dr. Zeng was honored by his international peers at the annual August dinner of the AAPM in Washington, DC.
Read more about the Low-Dose CT Grand Challenge here.