Related Experiment Video
Updated: Jan 18, 2026
13:44
Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns
Published on: August 30, 2013
43.6K
Artificial Intelligence for Assessment of Digital Mammography Positioning Reveals Persistent Challenges
Laurie R Margolies1, Georgia G Spear2,3, Jennifer I Payne4,5
1Department of Diagnostic, Molecular and Interventional Radiology, Icahn School of Medicine at Mount Sinai, Mount Sinai Health System, New York, NY, USA.
Journal of Breast Imaging
|May 30, 2025
Summary
Artificial intelligence (AI) identified common issues in mammography positioning quality across two health systems. This data can improve mammography screening through targeted education.
Area of Science:
- Radiology and Medical Imaging
- Artificial Intelligence in Healthcare
- Quality Improvement in Medical Diagnostics
Background:
- Mammographic breast cancer detection relies on optimal image quality, heavily influenced by patient positioning.
- Traditional assessment of mammography positioning quality (MPQ) is subjective and lacks standardization.
- Objective evaluation methods are needed to identify and quantify positioning deficiencies.
Purpose of the Study:
- To leverage artificial intelligence (AI) for objective evaluation of mammography positioning on digital screening mammograms.
- To identify and quantify specific unmet mammography positioning quality (MPQ) criteria.
- To compare the distribution and occurrence of unmet MPQ criteria between different healthcare systems.
Main Methods:
- Utilized a large dataset of 126,367 digital mammography studies (553,339 images).
- Developed and applied MPQ AI algorithms to assess unmet positioning criteria (e.g., PNL length, pectoralis muscle inclusion, image centering).
- Compared the similarity of unmet MPQ occurrence and rank order across two health systems.
Main Results:
- AI algorithms identified a substantial number of unmet MPQ criteria across both health systems.
- The most frequent unmet MPQ criteria included short posterior nipple line (PNL) length, inadequate pectoralis muscle, and excessive exaggeration on the craniocaudal (CC) view.
- No statistically significant difference was found in the rank order or probability distribution of unmet MPQ criteria between the two health systems (P = .844 and P = .92).
Conclusions:
- AI effectively identified a consistent pattern of unmet mammography positioning quality in routine clinical practice across two health systems.
- The findings highlight common areas of deficiency in mammography positioning.
- This objective data can inform the development of targeted educational strategies to enhance mammography quality and improve cancer detection rates.
