Using Five Machine Learning for Breast Cancer Biopsy Predictions Based on Mammographic Diagnosis

Abstract

Breast cancer is one of the causes of female death in the world. Mammography is commonly used for distinguishing malignant tumors from benign ones. In this research, a mammographic diagnosis method is presented for breast cancer biopsy outcome predictions using five machine learning which includes: Logistic Regression (LR), Linear Discriminant Analysis (LDA), Quadratic Discriminant Analysis (QDA), Radom Forest (RF) and Support Vector Machine (SVM) classification. The testing results showed that SVM learning classification performs better than other with accuracy of 95.8% in diagnosing both malignant an benign breast cancer, a sensitivity of 98.4% in diagnosing disease, a specificity of 94.4%. Furthermore, an estimated area of receiver operating characteristic (ROC) curve analysis for support vector machine (SVM) was 99.9% for breast cancer outcome predictions, outperformed the diagnostic accuracies of Logistic Regression (LR), Linear Discriminant Analysis (LDA), Quadratic Discriminant Analysis (QDA), Radom Forest (RF) methods. Therefore, Support Vector Machine (SVM) learning classification with mammography can provide highly accurate and consistent diagnoses in in distinguishing malignant and benign cases for breast cancer prediction