Related Experiment Video
Updated: Aug 8, 2026
13:44
Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns
Published on: August 30, 2013
Computer-aided mammographic screening for spiculated lesions
W P Kegelmeyer1, J M Pruneda, P D Bourland
1Sandia National Laboratories, Livermore, Calif.
Radiology
|May 1, 1994
Summary
Computer vision significantly improves mammogram screening. This AI tool enhanced radiologist detection of spiculated lesions, increasing cancer diagnosis accuracy without reducing specificity.
Area of Science:
- Radiology
- Medical Imaging
- Computer Vision
Background:
- Mammography is a key tool for breast cancer screening.
- Early detection of spiculated lesions is crucial for effective treatment.
- Radiologist performance can be enhanced with decision support tools.
Purpose of the Study:
- To evaluate a computer vision algorithm as a second reader for detecting spiculated lesions on screening mammograms.
- To assess the impact of computer-aided detection on radiologist performance.
Main Methods:
- An algorithmic computer process was developed for detecting spiculated lesions on digitized mammograms.
- The algorithm was tested on 85 clinical cases (36 with cancer, 49 negative).
- Four radiologists screened cases twice: once independently and once with computer-generated reports.
Main Results:
- The computer vision algorithm alone demonstrated 100% sensitivity and 82% specificity.
- Computer reports improved average radiologist sensitivity by 9.7% (from 80.6% to 90.3%).
- This improvement in sensitivity was statistically significant (P = .005) with no reduction in specificity.
Conclusions:
- Computer analysis of mammograms can substantially enhance radiologist screening efficacy.
- AI-powered second readers offer a promising approach to improve diagnostic accuracy in mammography.
- This technology has the potential to improve patient outcomes through earlier and more accurate cancer detection.
