Diagnosis of Mammographic Images for Breast Cancer Detection using FF-CSO Algorithm

Abstract

Many diseases in the human being are can be detected through different imaging technology. The mammography is one of the imaging technique through which breast cancer cells can be identified. The mammographic breast image is preprocessed for eradicating the pectoral muscle in the breast cancer identification that contains a mammogram for encircling the process of detection. The cancer tissues having higher pixel intensities can be detected easier than the remaining breast region. It is difficult to classify the breast tumor tissues into malignant and benign ones. The mammographic breast image is generally pre-processed to eliminate pectoral muscle for the optimal diagnosis. Further, the active contour based segmentation is done to separate out the cancer region from entire mammographic image acquired by special technique. Segmenting accurate region of cancer will help in classification so segmentation stage has its own importance. Two different optimization algorithms are merged to improve optimization accuracy. The algorithms considered are Firefly and chicken swarm optimization (FF-CSO) is used optimize feature set and optimize error function of DBN. Once segmentation is performed on the image, the next process is to extract features from that segmented region to collect useful information for distinguishing the cases. In this we extracted LBP features from the segmented images. The deep learning architectures is for classification.