Related Experiment Video
Updated: Jun 26, 2026
13:44
Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns
Published on: August 30, 2013
43.5K
Mammo-AGE: deep learning estimation of breast age from mammograms
Xin Wang1,2, Tao Tan3,4, Yuan Gao1,2,5
1Department of Radiology, The Netherlands Cancer Institute, Amsterdam, The Netherlands.
Nature Communications
|December 8, 2025
Summary
This study introduces a deep learning model to predict biological breast age from mammograms. This breast age estimation can help stratify cancer risk and potentially improve early detection and personalized screening strategies.
Area of Science:
- Radiology
- Artificial Intelligence
- Oncology
Background:
- Biological age is a key health indicator.
- Mammograms are crucial for breast cancer screening.
- Mammogram-based biological age prediction is underexplored.
Purpose of the Study:
- Develop a deep learning model to estimate breast biological age from mammograms.
- Assess the model's accuracy and correlation with chronological age.
- Investigate the association between breast age gap and breast cancer risk.
Main Methods:
- Developed a deep learning model using large mammogram datasets (95,826 images).
- Externally validated the model on additional datasets.
- Utilized occlusion analysis for model interpretation.
- Analyzed breast age gap in relation to breast cancer patients and future risk.
Main Results:
- Accurate breast age estimation (MAE: 4.2-6.1 years) with strong chronological age correlation.
- Predicted breast age effectively stratifies breast cancer risk.
- Higher breast age gaps observed in cancer patients and associated with increased future risk (HR: 1.013-1.022).
Conclusions:
- Mammogram-based biological breast age prediction is feasible and accurate.
- Breast age gap serves as a potential biomarker for breast health status and future risk.
- The model shows promise for early breast cancer detection and personalized screening.
