Related Experiment Videos
Computed assisted detection of interval breast cancers
K Moberg1, N Bjurstam, B Wilczek
1Karolinska Institutet, Department of Radiology, Stockholm Söder Hospital, Sodersjukhuset, 118 83 Stockholm, Sweden. kerstinb.mogerg@telia.com
European Journal of Radiology
|August 28, 2001
Summary
Computer-assisted detection (CAD) did not improve radiologist sensitivity for detecting interval cancers. However, CAD did not decrease radiologist specificity, suggesting potential for improved accuracy with enhanced CAD specificity.
Area of Science:
- Radiology
- Medical Imaging
- Cancer Detection
Background:
- Interval cancers are missed during screening mammography.
- Computer-assisted detection (CAD) systems aim to improve cancer detection rates.
Purpose of the Study:
- To evaluate the interval cancer detection rate using CAD.
- To assess CAD's impact on radiologists' sensitivity and specificity.
Main Methods:
- Three radiologists reviewed mammograms retrospectively.
- Interval cancer cases were mixed with normal screening mammograms.
- Radiologists interpreted cases with and without CAD assistance.
Main Results:
- Radiologists' sensitivity did not increase with CAD (17, 12, 11 without vs. 16, 10, 13 with CAD).
- Radiologists' specificity remained high and was not reduced by CAD (73-89% without vs. 78-92% with CAD).
- Non-mixed reading improved detection rates (21, 17, 19 interval cancers).
Conclusions:
- CAD alone showed high sensitivity but did not enhance radiologist sensitivity.
- Improving CAD specificity is crucial for its effective integration.
- CAD could potentially increase sensitivity when used as a double reader if specificity is improved.