Automatic Breast Tumor Diagnosis in MRI Based on a Hybrid CNN and Feature-Based Method Using Improved Deer Hunting...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Breast cancer diagnosis relies on early detection for effective treatment.
• Computer-aided diagnosis systems are crucial for automatic cancer detection.
• Existing methods require optimization for improved accuracy and efficiency.

Purpose of the Study:

• To develop an automatic breast tumor diagnosis system.
• To enhance breast cancer detection using a hybrid feature-based technique and an optimized convolutional neural network (CNN).
• To introduce a novel metaheuristic algorithm for optimization.

Main Methods:

• Utilized an improved Deer Hunting Optimization Algorithm (DHOA) for optimization.
• Employed a hybrid feature-based technique combined with an optimized CNN.
• Applied Haralick texture and Local Binary Pattern (LBP) for feature extraction.
• Implemented a preprocessing stage to simplify classification.

Main Results:

• The developed system achieved a high accuracy of 98.89% on the DCE-MRI dataset.
• The hybrid feature extraction and optimized CNN demonstrated significant efficiency.
• The novel metaheuristic algorithm contributed to improved diagnostic performance.

Conclusions:

• The proposed automatic breast tumor diagnosis system shows high potential and efficiency.
• The integration of DHOA, CNN, and advanced feature extraction offers a promising approach for breast cancer detection.
• Early and accurate diagnosis through such systems can improve patient outcomes and reduce healthcare costs.