Radiologists' perceptions of computer aided detection versus double reading for mammography interpretation

Area of Science:

• Radiology
• Medical Imaging
• Breast Cancer Screening

Background:

• Screening mammography interpretation relies on accurate detection of breast cancer.
• Computer-aided detection (CAD) and double reading are two methods to potentially improve mammography interpretation accuracy.
• Understanding radiologists' adoption and views on these methods is crucial for optimizing breast cancer screening.

Purpose of the Study:

• To investigate radiologists' current use of CAD and double reading in screening mammography.
• To assess radiologists' perceptions regarding the effectiveness and impact of CAD versus double reading.
• To identify factors associated with favorable perceptions and use of CAD.

Main Methods:

• A mailed survey was distributed to 257 radiologists within the Breast Cancer Surveillance Consortium.
• Latent class analysis was employed to categorize radiologists' perceptions of CAD and double reading.
• Data on usage patterns, perceived benefits (cancer detection rates, recall rates), and influencing factors were collected.

Main Results:

• 64% of radiologists used CAD for over half their screening mammograms; <5% used double reading extensively.
• More radiologists believed double reading improved cancer detection (74%) compared to CAD (55%).
• Favorable CAD perceptions correlated with perceived efficiency, academic affiliation, and fellowship training; favorable double reading perceptions were strongly linked to academic affiliation.

Conclusions:

• Radiologists generally hold more favorable perceptions of double reading by a human reader than by CAD, despite lower adoption rates.
• Both CAD and double reading are perceived favorably by most radiologists, albeit for distinct reasons.
• Perceptions and usage patterns highlight differing views on the value and implementation of these screening tools.