Computer Aided Detection for Early Detection of Breast Cancer

Abstract

According to World Health Organization breast cancer is leading cause of death newlineamong women worldwide. Mammogram processing is established as an effectual way newlinefor the early detection of breast cancers, which might significantly minimize the death newlinerate of breast cancer cases. There are numerous ways of determining the earlier newlinedetection of breast cancer. Amongst these methods, the mammography is considered newlineas the best screening technique for the earlier determination of breast cancer regions. However, the medical centre faces a major challenge in evaluating the high volume of newlinemammograms. The computer-aided detection is well-known due to its ability to newlineenhance the accuracy of earlier disease diagnosis. CAD systems utilize the computer newlinetechnologies for determining the abnormalities in the mammogram like masses, micro-calcifications, asymmetry and architecture distortion which plays a newlinefundamental role in the earlier detection of breast cancer and assist to minimize the newlinemortality rate amongst the women. This thesis presents three different techniques for early breast cancer detection using newlinethe CAD system. The first methodology presents mammogram classification using the newlinelongest line detection algorithm for pectoral muscle removal and decision tree newlineclassifiers. A total of 322 mammogram images evaluated for mammogram newlineclassification. In the second methodology mammogram classification and mass, detection is carried out with SVM and CNN classifier. A combination of DDSM and newlineMIAS database is used in second methodology. In the third methodology, mass is newlinesegmented and detected with Chaotic based chicken swarm optimization with deep newlineconvolutional neural network (Chaotic-CSO-based DCNN) classifier. A total of 900 newlinemammograms from DDSM database are evaluated for this technique. These newlinetechniques resulted in improved accuracy in the classification of mammograms as newlinenormal and abnormal for early detection of breast cancer. newline