DefinePK

Abstract

Ultra wide-band (UWB) microwave technology is a promising candidate to detect the early breast cancer. This paper aims to depict pattern recognition performance of support vector machine (SVM) for confocal UWB breast tumor imaging dataset. A novel feature extraction technique is also introduced in this paper for the signal classification perfectly and promptly. SVM classifier functions the comparative study between SVM kernel functions includes linear function, radial basis function, polynomial and multi layer perceptions are investigated and verified for pattern recognition performance with the help of receiver operating characteristic (ROC) graph and confusion matrix. The main motto of this paper is to identify the tumor in its smallest dimension from available works including their data using the proposed feature extraction. In total, thirteen different sizes of benign tumors are being considered where the smallest and largest tumor sizes utilized are 1mm and 9 mm, respectively.