Improved Breast Cancer Detection in Mammography Images

Abstract

Cancer, characterized by uncontrolled cell division, is an incurable ailment, with breast cancer being the most prevalent form globally. Early detection remains critical in reducing mortality rates. Medical imaging is vital for localizing and diagnosing breast cancer, providing key insights for identification. This study introduces an automatic hybrid feature recognition method for breast cancer diagnosis using images from two mammography datasets. The method employs a convolutional neural network (CNN) and local binary pattern (LBP) for feature extraction. Correlation-based feature selection techniques reduce dimensionality, enabling faster computation and improved accuracy. The proposed model's superiority is established through comparative analysis with cutting-edge deep models, achieving 96% accuracy across the MIAS and INbreast datasets. The hybrid method demonstrates high accuracy with minimal computational tasks.