SVM-based classification of breast tumour phantoms using a UWB radar prototype system

In this paper, a follow-up study exploring the classification of phantoms mimicking benign and malignant breast tumours, using a pre-clinical Ultra Wideband (UWB) prototype imaging system, is presented. A database of 13 benign and 13 malignant tumour phantoms was created using material which mimicked the dielectric properties of tumour tissues in the 1-6GHz frequency range. The classification was performed using a machine learning algorithm - Support Vector Machines (SVM) - and the results were compared to those of a previous study by the authors where Linear Discriminant Analysis and Quadratic Discriminant Analysis were considered.