Related Experiment Videos
Deep meta-learning framework with U-Net segmentation and optimization-based feature selection for breast cancer
Balasubramaniam Pudhupalayam Marimuthu1, Paulraj Ranjith Kumar2, Arivoli Sundaramurthy3
1Hindusthan Institute of Technology, Coimbatore, Tamilnadu, India.
Quantitative Imaging in Medicine and Surgery
|August 12, 2026
Summary
This study introduces an efficient deep meta-learning framework for accurate breast cancer detection using MRI images. The novel approach significantly improves diagnostic accuracy and reduces false positives, aiding early clinical decisions.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- Breast cancer is a leading global health challenge, necessitating early detection for improved patient survival rates.
- Current medical imaging techniques for breast cancer detection have limitations, including high computational costs and inaccuracies.
- There is a critical need for advanced, automated systems for reliable breast cancer diagnosis.
Purpose of the Study:
- To develop an efficient deep meta-learning framework for accurate breast cancer detection and classification using MRI images.
- To enhance diagnostic accuracy and reduce false-positive rates in computer-aided diagnosis (CADx) systems.
- To support early clinical decision-making through improved automated diagnostic capabilities.
Main Methods:
- The framework integrates Adaptive Median Filtering (AMF) for preprocessing, U-Net for tumor segmentation, and Growth Distribution Depth (GDD) for feature extraction.
- Salp Swarm Optimization (SSO) was used for feature selection, followed by deep meta-learning classification using fine-tuned CNN models (AlexNet, VGG16, LeNet).
Main Results:
- The proposed framework demonstrated superior performance over existing methods for breast cancer detection from MRI scans.
- The VGG16-based classifier achieved the highest accuracy (98.82%), precision (97.1%), recall (99.11%), and F1-score (98.25%).
- Integrated tumor segmentation and optimization-based feature selection significantly boosted classification accuracy and reduced false detections.
Conclusions:
- The developed deep meta-learning framework offers an effective and reliable automated method for breast cancer detection using MRI.
- The system enhances diagnostic accuracy and efficiency, showing potential for integration into computer-aided diagnosis systems.
- This approach can assist clinicians in achieving earlier and more precise breast cancer detection.