CAD on-demand improves reader interpretation

Tools

Radiologists have more success in finding malignant breast masses with computer-aided detection marks when they are hidden until needed, rather than being continually on display, according to an article published online in Radiology.

In the article, the authors--led by Rianne Hupse, of Radboud University Nijmegen Medical Center in Nijmegen, the Netherlands--noted that the use of CAD can lead to disappointing results because of incorrect reader interpretations. They also referred to research that showed that reader performance could improve by combining user scores with CAD marks.

According to an AuntMinnie.com article, this inspired the researchers to develop a CAD system in which radiologists, instead of using the technology to help in the initial detection of suspicious regions, used it to interpret those regions. In that kind of system, CAD marks are displayed on-demand for regions with a suspiciousness score so that the reader doesn't rely on CAD to detect masses, but instead to improve interpretation.

In the study, nine screening radiologists read 200 studies--63 containing at least one-screen-detected mass, 17 false-negative studies, 20 false-positive studies and 100 normal studies--once with CAD prompts and once interactively (when the CAD marks and associated scores were hidden until queried by the reader). The authors found that reader performance increased when CAD results were displayed interactively.

According to AuntMinnie.com, readers using CAD in the prompting mode had a sensitivity of 84 percent (67 of 80 cases), while they had a sensitivity of 91 percent (73 of 80 cases) in the interactive mode.

"When used as decision support, interactive use of CAD for malignant masses on mammograms may be more effective than the current use of CAD," the authors concluded.

For more:
- see the article abstract in Radiology
- read the article in AuntMinnie.com

Related Articles:
Computer-aided detection could hinder image interpretation
CAD software reduces variability for radiologists reading images
FDA publishes final computer-assisted detection technology guidance
Computer-aided detection aids cancer testing, treatment