Innovative deep learning classifiers for breast cancer detection through hybrid feature extraction techniques.

Area of Science:

• Medical Imaging
• Artificial Intelligence in Healthcare
• Oncology

Background:

• Breast cancer is a leading cause of death in women, making early detection crucial for survival.
• Accurate mammogram analysis is essential for effective breast cancer screening and diagnosis.

Purpose of the Study:

• To develop and evaluate a hybrid classification method for enhanced mammogram analysis.
• To combine handcrafted statistical features with deep learning for improved breast cancer detection.

Main Methods:

• Mammogram preprocessing using Shearlet Transform.
• Image segmentation via Improved Otsu thresholding and Canny edge detection.
• Feature extraction using GLCM, GLRLM, and 1st-order statistics, fed into a 2D BiLSTM-CNN model.

Main Results:

• The hybrid method achieved 97.14% accuracy on the MIAS dataset.
• Outperformed several benchmark models in mammogram classification.
• Demonstrated the effectiveness of combining statistical and deep learning features.

Conclusions:

• The proposed hybrid approach significantly improves breast cancer classification performance.
• This method shows potential to assist radiologists in more effective breast cancer screening.
• The integration of handcrafted features and deep learning offers a promising direction for medical image analysis.