Related Experiment Video
Updated: Jun 26, 2026
05:28
Clinical Imaging of Microwave Mammography
Published on: November 14, 2025
Machine Learning-Based Classification of BI-RADS 4 and BI-RADS 5 Microcalcifications in Mammography Combined with
Sevgi Ünal1, Enes Açıkgözoğlu2
1Department of Radiology, Izmir Katip Celebi University Ataturk Training and Research Hospital, Izmir 35360, Turkey.
Tomography (Ann Arbor, Mich.)
|June 25, 2026
Summary
Machine learning models integrating mammographic features and DCE-MRI data show promise for classifying breast microcalcifications. Logistic Regression achieved high accuracy, aiding in differentiating benign from malignant lesions.
Area of Science:
- Radiology and Medical Imaging
- Machine Learning in Healthcare
- Oncology
Background:
- Breast cancer is a leading cause of mortality in women globally.
- Accurate characterization of mammographic microcalcifications is crucial for diagnosis.
- Differentiating benign from malignant microcalcifications (BI-RADS 4-5) is challenging due to overlapping patterns.
Purpose of the Study:
- Develop a machine learning model to predict pathological diagnosis of breast microcalcifications.
- Integrate mammographic descriptors, patient age, and DCE-MRI findings.
Main Methods:
- Utilized a dataset of 53 biopsy-confirmed cases.
- Evaluated multiple machine learning algorithms (Logistic Regression, SVM, KNN, etc.).
- Performed hyperparameter optimization using grid search and cross-validation.
Main Results:
- Logistic Regression demonstrated the highest performance (accuracy 0.909, F1-score 0.889).
- AdaBoost achieved perfect recall (1.000) in internal evaluation.
- Results are preliminary due to limited sample size and lack of external validation.
Conclusions:
- Structured radiological descriptors and DCE-MRI data may aid malignancy risk stratification for BI-RADS 4-5 microcalcifications.
- Further multicenter studies are needed for clinical implementation.
