Early Breast Cancer Diagnosis and Risk Prediction based on Machine Learning

Abstract

Breast cancer is a disease in which life-threatening (malignant) cells in the breast multiply out of hand, making it the second most fatal type of cancer in women widely. Hence, to diminish the mortality rate and increasing the chances of survival, it is crucial to uncover it as early as attainable. This paper focused on comparing the different classifiers which are support vector machine, naïve Bayes, and K-nearest neighbor algorithms using the DDSM dataset. The target of this computer-aided system is to combine these classification techniques with image pre-processing methods so to compare their performance to find out the most satisfactory approach. The crux is to use the advantages of these techniques to obtain maximum optimal performance. For the comparative study, the digital mammogram of the breast is passed to histogram equalization for image pre-processing which enhances the necessary feature while removing noise that is present in the mammogram, the refined mammograph is then passed to wavelet transformation to extract all the important features for the classification.