An Effective Feature Extraction Based Particle Swarm Optimization with Support Vector Machine for Biomedical Mammogram Image Diagnosis

摘要

Breast cancer is a second significant reason for the increased mortality rate of women in both developing and developed countries. When abnormalities in breast cancer are identified in the earlier stage, there is a greater chance to increase the survival rate. This paper presents a new breast cancer diagnosis model using feature extraction and classification process. The presented model involves preprocessing, Hough transforms based feature extraction, particle swarm optimization (PSO) with support vector machine (SVM) called PSO-SVM based classification. Initially, preprocessing takes place to remove the noise present in the image. Then, Hough transform based feature extraction process is carried out to extract the features exist in the image. Then, the PSO-SVM model is applied to classify breast cancer images into normal and abnormal. The validation of the presented PSO-SVM model takes place on MIAS dataset and the experimental outcome indicated that the presented model achieved a maximum accuracy of 94.61%.