Related Experiment Video
Updated: Jul 18, 2026
13:44
Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns
Published on: August 30, 2013
Testing the effect of computer-assisted detection on interpretive performance in screening mammography
Stephen H Taplin1, Carolyn M Rutter, Constance D Lehman
1Group Health Cooperative, Center for Health Studies, Seattle, WA 98101, USA.
AJR. American Journal of Roentgenology
|November 23, 2006
Summary
Computer-assisted detection (CAD) improved specificity in breast cancer screening but did not significantly alter overall sensitivity. CAD
Area of Science:
- Radiology
- Medical Imaging
- Oncology
Background:
- Breast cancer screening relies on mammography, with interpretation challenges in dense breast tissue.
- Computer-assisted detection (CAD) systems aim to improve accuracy by highlighting potential abnormalities.
Purpose of the Study:
- To evaluate if CAD enhances breast cancer detection sensitivity without compromising specificity.
- To assess CAD's impact on interpreting fatty versus dense breast tissue.
Main Methods:
- A stratified random sample of 341 screening mammograms was used.
- 19 radiologists interpreted mammograms with and without CAD in separate test settings.
- Sensitivity and specificity were compared using regression models.
Main Results:
- CAD did not change overall sensitivity but improved specificity (72% vs. 75%).
- CAD improved sensitivity for lesions it marked but decreased it for unmarked lesions.
- Performance differences were more pronounced in radiologists interpreting >50 mammograms/week.
Conclusions:
- CAD enhances interpretive specificity in mammography but does not improve overall sensitivity.
- The system's effectiveness is influenced by its ability to mark lesions accurately.
- Breast density did not impact CAD's performance.
