Related Experiment Video
Updated: Oct 4, 2025
07:15
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
7.0K
Deep learning model improves radiologists' performance in detection and classification of breast lesions
Yingshi Sun1, Yuhong Qu1,2, Dong Wang3
1Key Laboratory of Carcinogenesis and Translational Research (Ministry of Education/Beijing), Department of Radiology, Peking University Cancer Hospital & Institute, Beijing 100142, China.
Chinese Journal of Cancer Research = Chung-Kuo Yen Cheng Yen Chiu
|February 7, 2022
Summary
This study developed an artificial intelligence (AI) model for mammography, significantly improving breast lesion detection and diagnosis accuracy. The AI tool enhanced radiologist performance and reduced reading time, showing great clinical potential.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Deep Learning
Background:
- Deep learning algorithms show promise in computer-aided diagnosis for mammography.
- Large-scale clinical applications of AI in mammography are currently limited.
- There is a need to develop and validate AI models for mammographic analysis.
Purpose of the Study:
- To develop and validate an artificial intelligence (AI) model for mammography.
- To assess the impact of the AI model on radiologists' diagnostic performance.
- To evaluate the efficiency of AI-assisted mammogram interpretation.
Main Methods:
- Retrospective collection and randomization of mammograms from six centers for model training and validation.
- Comparison of 12 radiologists' performance with and without the AI model.
- Prospective evaluation of the AI model in diagnosing breast lesions using FROC and ROC curves.
Main Results:
- The AI model achieved high sensitivity (0.908) for lesion detection at a 0.25 false positive rate.
- AI-assisted interpretation by radiologists improved diagnostic accuracy (AUC 0.852 vs. 0.805) and reduced reading time.
- Prospective application demonstrated excellent performance with AUC of 0.983 and high sensitivity (94.36%) and specificity (98.07%).
Conclusions:
- The developed AI model demonstrates high accuracy in detecting and diagnosing breast lesions.
- AI integration significantly enhances radiologists' diagnostic capabilities and efficiency.
- The AI model offers a valuable tool for improving breast cancer screening and diagnosis.
