SVM-Based Microwave Breast Tumour Classification

This paper presents the feasibility of the benign and malignant breast tumor classification according to dielectric properties. The experiment data are obtained by simulating the human breast with the help of COMSOL Multiphysics, in which way electric field amplitude of different samples is collected. Support Vector Machines (SVM) is utilized to analyze the microwave signal and distinguish benign and malignant breast tumors. The capability and reliability of classification are compared with Decision Tree and Random Forest.