Classification of Mammogram Images by Using SVM and KNN

Abstract

Breast cancer is a fairly diverse illness that affects a large percentage of women in the west. A mammogram is an X-ray-based evaluation of a woman's breasts to see if she has cancer. One of the earliest prescreening diagnostic procedures for breast cancer is mammography. It is well known that breast cancer recovery rates are significantly increased by early identification. Mammogram analysis is typically delegated to skilled radiologists at medical facilities. Human mistake, however, is always a possibility. Fatigue of the observer can commonly lead to errors, resulting in intraobserver and interobserver variances. The image quality affects the sensitivity of mammographic screening as well. The goal of developing automated techniques for detection and grading of breast cancer images is to reduce various types of variability and standardize diagnostic procedures. The classification of breast cancer images into benign (tumor increasing, but not harmful) and malignant (cannot be managed, it causes death) classes using a two-way classification algorithm is shown in this study. The two-way classification data mining algorithms are utilized because there are not many abnormal mammograms. The first classification algorithm, k-means, divides a given dataset into a predetermined number of clusters. Support Vector Machine (SVM), a second classification algorithm, is used to identify the optimal classification function to separate members of the two classes in the training data