Related Experiment Video
Updated: May 19, 2026
08:27
Image Recognition and Parameter Analysis of Concrete Vibration State Based on Support Vector Machine
Published on: January 5, 2024
Combining support vector machine with genetic algorithm to classify ultrasound breast tumor images
Wen-Jie Wu1, Shih-Wei Lin, Woo Kyung Moon
1Department of Information Management, Chang Gung University, Tao-Yuan, Taiwan. wjwu@mail.cgu.edu.tw
Summary
This study introduces a computer-aided diagnosis system for breast tumor ultrasound images, achieving 95.24% accuracy. The approach significantly reduces feature extraction and model optimization time, aiding physicians in diagnosis.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- Accurate classification of breast tumors is crucial for timely treatment.
- Computer-aided diagnosis (CAD) systems aim to improve diagnostic accuracy and efficiency.
- Optimizing feature selection and model parameters is key to CAD system performance.
Purpose of the Study:
- To enhance the classification accuracy of ultrasound breast tumor images.
- To reduce the computational time for feature extraction and model optimization in CAD systems.
- To develop a clinically useful CAD system for differentiating benign from malignant breast tumors.
Main Methods:
- Automatic segmentation of ultrasound breast tumors using a level set method.
- Extraction of auto-covariance texture and morphologic features.
- Simultaneous feature selection and support vector machine (SVM) parameter optimization using a genetic algorithm.
Main Results:
- Achieved a high classification accuracy of 95.24% for differentiating benign from malignant breast tumors.
- Reduced feature extraction computing time to 8% compared to systems without feature selection.
- Significantly decreased the time required for finding optimal classification model parameters compared to grid search.
Conclusions:
- The proposed CAD system demonstrates high accuracy and efficiency in breast tumor classification.
- The integrated approach of feature selection and parameter optimization offers clinical utility.
- The system can aid in reducing unnecessary biopsies and assist physicians in avoiding misdiagnosis.