New Approaches for Ultrasonic Breast Imaging

摘要

The aim of this study is to estimate sound velocity and attenuation coefficient of a region of interest in the tissue in order to discriminate breast cancer from benign tissues. The reconstruction method with a metal reflector for sound velocity and attenuation coefficient distribution were proposed and tested on tissue-mimicking phantoms. According to our clinical results, carcinoma could be discriminated from fibroadenoma and fat by choosing two appropriate thresholds for the relative sound velocity, 18.5 m/s and -28.3 m/s. Besides, we evaluated the relationships between texture, morphological, and acoustic features in computer-aided diagnosis, and assessed the effects of combining different feature sets. Both texture and morphological features were extracted from B-mode images. Acoustic features were estimated based on our reconstruction method. Generally, the correlation coefficients for features within the texture feature group were relatively high (0.48-0.79), whereas those between different feature categories were relatively low (0.17-0.40). This suggests that combining different sets of features would improve the computer-aided diagnosis of breast cancer.