Fusion of Wavelet and Morphological Features for Breast Cancer Diagnosis in Ultrasound Images

Cancer remains one of the major concerns of deaths worldwide. Early detection is the key point in reducing the cancer mortality. Automatic systems are needed to assist radiologists in the cancer detection and diagnosis. Hence, there are strong needs for the development of computer aided diagnosis (CAD) systems which have the capability to help radiologists in decision making. The aim of this work is to develop a computer aided system for breast cancer diagnosis in ultrasound images. The developed system consists of segmentation, feature extraction, feature selection and classification. The marker controlled watershed technique is used to segment the region of interest (ROI). In the feature extraction step, the wavelet transform is applied then the texture and statistical features of ROI are extracted. In addition, a set of morphological features are extracted directly from ROI in spatial domain. The obtained features are combined together to produce the feature vector. In order to select the most discriminative feature, a feature ranking technique is used to determine the capability of each feature. In the classification step, support vector machine (SVM), classification and regression trees (CART) and classification rule classifiers are used to classify the ROI as benign or malignant. The proposed method is validated using 10 fold cross-validation. The results show that classification rule classifier outperforms SVM and CART classifiers.