Early Detection of Breast Cancer Using Support Vector Machine With Sequential Minimal Optimization

Abstract

Breast cancer is largely occurring cancer disease and one of the leading causes of death among the women in the world. Studies have shown that early detection can bring down the mortality rate significantly. Mammography is the most popular cancer detection technique but it is painful as well as it uses X-ray to capture images which spreads radiation in the body. Radiation is one of the causes of breast cancer. Thermography is a method of detection which is noninvasive, painless, and radiation-free. By seeing these thermal images, doctors can’t say whether the patient is having cancer or not that’s why they use computer-aided diagnosis (CAD) to see the status by feeding these thermal images to model. The detection of breast cancer from these data is very crucial and needs some sophisticated image processing and machine learning techniques. Modern machine learning technique has become a popular tool for the diagnosis of breast cancer. Among various methods, support vector machine (SVM) classification is one of the most popular supervised learning methods. Performance of SVM classification by using sequential minimal optimization (SMO) algorithm is evaluated and our proposed model is giving better result in terms of accuracy (94.6%), recall (89.5%), and execution time (0.085 s) on Wisconsin data set.