Performance Analysis of Various Classification Algorithms for Computer-Aided Breast Cancer Diagnosis System Using Thermal Medical Images

Abstract

One of the most widespread cancer types is breast cancer all over the world. It affects both women and men. Detection of cancer in early-stage is very critical in terms of treatment success. Many studies have been done in image processing, for the detection of cancer using computer-aided diagnosis systems. In this study, the performance of various classification algorithms in cancer detection was analyzed on a thermal image dataset. For this purpose, a graphical user interface based system was developed using MATLAB. The developed system uses five different algorithms; Decision Tree, Support Vector Machine (SVM), Logistic Regression Analysis, K Nearest Neighborhood (KNN), Linear Discriminant Analysis. According to the obtained results, KNN and SVM provide the best performance. The developed system can be used as an assistant system to produce an objective result for the expert in breast cancer diagnosis with the %98.8 success rate.