• Center on Health Equity and Access
  • Clinical
  • Health Care Cost
  • Health Care Delivery
  • Insurance
  • Policy
  • Technology
  • Value-Based Care

Translating a Radiologist's Visual Expertise Into Improving CAD

Article

Scientists at Brigham and Women’s Hospital are evaluating the depth of a trained radiologist’s eye to detect abnormalities on a mammogram, in an attempt to improve screening and earlier detection of disease.

Scientists at the University of York, in colabortion with the Visual Attention Laboratory at Brigham and Women’s Hospital, are evaluating the depth of a trained radiologist’s eye to detect abnormalities on a mammogram, in an attempt to improve screening and earlier detection of disease. The researchers found that while radiologists are adept at detecting a global signal of abnormality, their ability to precisely localize the abnormality is based on chance.

,” Jeremy Wolfe, PhD, co-author on the study, said in a statement. "

“Radiologists can have ‘hunches’ after a first look at a mammogramWe found that these hunches are based on something real in the images. It’s really striking that in the blink of an eye, an expert can pick up on something about that mammogram that indicates abnormality."

Wolfe added that radiologists also have the knack to sometimes detect abnormalities in the other breast that has not developed a lesion.

Based on their previous results, the authors knew that radiologists can distinguish between normal and abnormal mammograms after a half-second glimpse. Although they are more thorough in the clinic, when using computer-assisted diagnosis, or CAD, their ability to quickly grasp abnormalities means there are signs that are being rapidly picked by the expert’s eye. To take this a step further, for their current study, the researchers tested whether breast tissue symmetry, breast density, image size, resolution, or other characteristics were contributing factors to the success rate of the radiologists.

"Radiologists often have ‘hunches’ about images on first glimpse. Our work shows that those hunches are based on something real in the image. Looking to the future, any medical field that requires image screening and diagnostics, such as dermatologists and pathologists, we think might use analogous signals," said lead author Karla Evans, PhD, University of York, in a statement.

The data suggested that the texture of the breast tissue could significantly weigh in on accurate diagnosis by radiologists. "

Although this signal is not, in itself, definitive, it has the potential to be used in automated aids to medical screening and incorporated into training protocols for medical experts, speeding up and improving cancer detection," the authors concluded.

“Together, these results suggest that radiologists may be picking up on some sort of early, global signal of abnormality that is unknown to us at this point,” said Wolfe. The team plans to explore ways to define this early “signal” so it can be incorporated into improving CAD systems used in screening tests.

The study was conducted in collaboration with the University of York and Leeds in the United Kingdom and MD Anderson Cancer Center in Texas.

Reference

PNAS

Evans KK, Haygood TM, Cooper J, Culpan A, Wolfe J. A half-second glimpse often lets radiologists identify breast cancer cases even when viewing the mammogram of the opposite breast [published online August 29, 2016]. . doi: 10.1073/pnas.1606187113.

Related Videos
Video 1 - "Diagnosing and Understanding the Pathogenesis of Bronchiectasis"
Video 4 - "Challenges in Autoantibody Screening for Type 1 Diabetes"
Jeff Stark, MD, vice president, head of medical immunology, UCB
Video 7 - "Prior Authorization and Access to Targeted Treatment for Ph+ ALL Patients"
Video 7 - "Prior Authorization and Access to Targeted Treatment for Ph+ ALL Patients"
Video 6 - "Community Partnership: Increasing Public Awareness of CVD"
Video 6 - "Community Partnership: Increasing Public Awareness of CVD"
Screenshot of Raajit Rampal, MD, PhD
 Laura Ferris, MD, PhD, professor of dermatology, University of Pittsburgh
Related Content
© 2024 MJH Life Sciences
AJMC®
All rights reserved.