Related Experiment Video
Updated: Jun 10, 2026
13:44
Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns
Published on: August 30, 2013
Diagnosing breast masses in digital mammography using feature selection and ensemble methods.
1National Yunlin University of Science and Technology, Yunlin, Taiwan. g9521806@yuntech.edu.tw
Journal of Medical Systems
|August 13, 2010
Summary
Accurate breast cancer prediction is crucial. This study found that feature selection improves prediction accuracy, and ensemble classifiers outperform single models for breast cancer diagnosis.
Area of Science:
- Medical Imaging
- Machine Learning
- Biostatistics
Background:
- Accurate breast cancer prediction is essential for effective treatment.
- Current prediction methods require refinement for improved accuracy.
Purpose of the Study:
- To enhance breast cancer prediction accuracy using feature selection and machine learning models.
- To evaluate the effectiveness of feature reduction techniques and ensemble classifiers.
Main Methods:
- Applied forward selection (FS) and backward selection (BS) for feature reduction.
- Utilized decision tree (DT) and support vector machine-sequential minimal optimization (SVM-SMO) classifiers.
- Developed ensemble methods combining DT and SVM-SMO for improved diagnostic performance.
Main Results:
- Feature reduction significantly improved predictive accuracy in breast cancer detection.
- The 'density' feature was identified as irrelevant in the analyzed mammogram dataset.
- Ensemble classifiers demonstrated superior accuracy compared to individual classifiers for breast cancer diagnosis.
Conclusions:
- Feature selection is a valuable technique for optimizing breast cancer prediction models.
- Ensemble machine learning approaches offer enhanced diagnostic performance for breast cancer.
- The study highlights the importance of data preprocessing and model selection in medical diagnostics.
