Classification of Breast Cancer From Electrical Impedance Measurements Dataset in Samples of Freshly Excised Breast Tissues

Abstract

The breast cancerous issues are increasing day by day for various reasons and even it can be found in young women. Early detection decreases the death rate and reduces painful medical treatments such as surgery and chemotherapy. Electrical Impedance Spectroscopy is a powerful and painless, low-cost detection technique, and it can be used with Mammography and MRI scans. The paper analyses the EIS dataset for classifying freshly exercised breast tissues using four diferent machine learning algorithms: Support Vector Machine, Decision Tree, Random Forest and Modifed Random Forest. The results are verifed with ANOVA statistical models on four diferent accuracy results of six classes. The results proved that Modifed Random Forest (MRF) works best as providing an accuracy mean value as 99% on 106 datasets with 15% as testing size during the training phase. The one way ANOVA results are also proved that standard error of 0.005, the signifcance of 0.003 and covariance of 1.14 for MRF.