Researchers will compete to create high-quality images from low-dose CT patient data

Science Highlights
December 1, 2015

Researchers who specialize in creating algorithms for reconstructing or denoising CT images are being challenged to see who can produce the best quality images from a shared set of low-dose patient CT data. The goal of the challenge is to accelerate the development of new methods that will enable the radiation dose of CT scans to be reduced without compromising image quality.

The challenge was announced on November 29 at the 2015 annual meeting of the Radiological Society of North America by Cynthia McCollough, Ph.D., director of the Mayo Clinic’s CT Clinical Innovation Center.

Since their inception, CT scans have led to significant improvements in the diagnosis and treatment of numerous medical conditions and diseases. However, CT scanners expose patients to ionizing radiation, which at high doses may elevate individuals’ lifetime risk of developing cancer or other health problems.

In 2012, NIBIB put out a call for researchers to submit groundbreaking ideas that would help radically decrease the amount of radiation used in CT scans, while maintaining the high quality of the images produced. McCollough answered the call and was funded by NIBIB to develop data sets, metrics, and software tools that could be disseminated to the research community so that it could more easily create and compare new approaches to reduce the radiation dose of routine CT scans while maintaining quality.

Over the past three years, McCollough and colleagues have created a library of raw data from patient CT scans that can be disseminated to the research community in a vendor neutral format.

Now, McCollough is hoping to expand the community of researchers working on reducing dose in CT imaging by inviting the medical imaging, computer science, electrical engineering, nondestructive testing, and Homeland Security CT communities to participate in a grand challenge.

“We are excited to challenge this broad pool of experts to produce the best images they can from these common data sets in hopes of identifying the most promising methodologies for reducing noise, and hence dose, in CT imaging,” said McCollough. 

In early January, challenge participants will receive data sets from the library that have been manipulated to reflect reduced radiation dose. Participants will have four months to create new or modify existing reconstruction or denoising techniques using the provided data.

Radiologists from the Mayo Clinic will perform a blinded and randomized interpretation of the images with the specific task of detecting liver lesions.

The three highest performing participants will be invited to present their algorithm and results at the 2016 annual meeting of the American Association of Physicists in Medicine (AAPM) in a Low-Dose CT Grand Challenge Scientific Symposium.

Information about how to register and additional details about the contest and awards is available on the AAPM website.