Related Experiment Video
Updated: Aug 29, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.9K
Breast MRI Tumor Automatic Segmentation and Triple-Negative Breast Cancer Discrimination Algorithm Based on Deep
Ying-Ying Guo1, Yin-Hui Huang2, Yi Wang1
1Department of CT/MRI, The Second Affiliated Hospital of Fujian Medical University, Quanzhou 362000, China.
Computational and Mathematical Methods in Medicine
|September 12, 2022
Summary
This study introduces a novel CNN-SVM network for automated breast tumor segmentation in MRI scans. The method accurately segments tumors, showing promise for early detection and treatment planning in breast cancer.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- Breast cancer incidence is rising, particularly in younger demographics.
- Magnetic Resonance Imaging (MRI) is crucial for breast tumor detection and treatment planning.
- Automated segmentation methods are needed due to increasing complexity and time constraints of manual segmentation.
Purpose of the Study:
- To develop and evaluate an automated segmentation method for breast tumors using MRI.
- To improve the accuracy and efficiency of breast tumor segmentation.
Main Methods:
- A combined Convolutional Neural Network (CNN) and Support Vector Machine (SVM) network was proposed.
- The CNN extracts features, and the SVM classifies these features for segmentation.
- The network was trained and tested on a collected breast tumor dataset.
Main Results:
- The proposed CNN-SVM network achieved high performance metrics: 0.93 (DSC coefficient), 0.95 (PPV), and 0.92 (sensitivity).
- Comparative analysis showed superior performance over existing segmentation frameworks.
- The method demonstrated accurate and efficient segmentation of breast tumors.
Conclusions:
- The CNN-SVM network provides effective breast tumor segmentation from MRI data.
- The method adapts to variations in breast tumors, ensuring accurate and efficient segmentation.
- This approach holds significant potential for future identification of triple-negative breast cancer.
