Multi-feature Fusion for Ultrasound Breast Image Classification of Benign and Malignant

Abstract

Computer-aided diagnosis (CAD) system for breast ultrasound can effectively assist doctors in classifying benign and malignant of breast tumors. CAD system increases the work efficiency and diagnostic accuracy, and it also reduces the rate of misdiagnosis. This paper proposes a method based on a histogram of oriented gradients (HOG), local binary pattern (LBP) and gray-level co-occurrence matrix (GLCM) feature extraction combined with machine learning classifier – support vector machine (SVM) to classify benign and malignant breast tumor ultrasound images. Besides, due to the lack of the latest and comprehensive ultrasonic image dataset, this work collected a latest ultrasound image dataset, which considers 350 benign lesions and 257 malignant tumors proven by biopsy. The ability of the classifier to predict the pathological classes is measured by the receiver operating characteristic (ROC) curve and obtain good classification result.