Daisuke Fukuoka, Tomoko Matsubara
Ultrasound Imaging: Classification Methods for Masses
Breast cancer is a common cancer among women all over the world. Combined with mammography, ultrasonography is used to suggest the degree of suspicion of malignancy to help determine whether follow-up or biopsy are necessary. Sonographic features that are assessed include shape, margin, echotexture, posterior acoustic shadowing, and orientation.
Recently, various classification methods using shape and other ultrasonographic features have been reported. Horsch et al.  reported a CAD method that is based on the automatic segmentation of lesions and the automatic extraction of four features related to the lesion: shape, margin, texture, and posterior acoustic behavior. Chen et al.  reported on methodology based on fractal analysis and k-means clustering. Other classification methods have been proposed that use the statistical properties of echo signals [100, 253, 254, 278, 291, 292]. K-distribution and Nakagami distribution are used for modeling the echo signals. Takemura et al.  compiled a total of 208 features for discrimination, including those based on a parameter of a log-compressed K-distribution. Their proposed system classifies types of diseases as cancer, cyst, or fibroadenoma (a common benign entity) using an ensemble classifier based on the AdaBoost algorithm with feature selection.
Tsui et al.  examined five contour feature parameters (tumor circularity, standard deviation of the normalized radial length, area ratio, roughness index, and standard deviation of the shortest distance) and calculated the Nakagami parameters estimated from the ultrasonic backscattered signals. The Nakagami parameters are only dependent on the statistical distribution of the echo waveform and are not affected by the echo amplitude. In another paper, Tsui et al.  suggested that the Nakagami image can visualize the scattering properties of breast lesions. Figure 3.59 shows the B-mode image and the corresponding Nakagami image of a benign breast tumor.