Computational and Mathematical Methods in Medicine / 2016 / Article / Tab 2 / Research Article
A Fusion-Based Approach for Breast Ultrasound Image Classification Using Multiple-ROI Texture and Morphological Analyses Table 2 The morphological and texture features employed for tumor classification.
Category Feature Code Description Texture Autocorrelation [30 ] TF1 Twenty texture features (TF1–TF20) are extracted from GLCM matrices computed using four distances ( pixels) and four orientations (θ = 0°, 45°, 90°, 135°) Contrast [12 ] TF2 Correlation [30 ] TF3 Cluster prominence [30 ] TF4 Cluster shade [30 ] TF5 Dissimilarity [30 ] TF6 Energy [30 ] TF7 Entropy [30 ] TF8 Homogeneity [30 ] TF9 Maximum probability [30 ] TF10 Sum of squares [27 ] TF11 Sum average [27 ] TF12 Sum entropy [27 ] TF13 Sum variance [27 ] TF14 Difference variance [27 ] TF15 Difference entropy [27 ] TF16 Information measure of correlation I [27 ] TF17 Information measure of correlation II [27 ] TF18 Inverse difference normalized [31 ] TF19 Inverse difference moment normalized [31 ] TF20 Morphological Tumor area [20 ] MF1 Ten morphological features (MF1–MF10) are extracted directly from the tumor Perimeter [20 ] MF2 Form factor [13 , 17 ] MF3 Roundness [13 , 17 ] MF4 Aspect ratio [13 , 17 ] MF5 Convexity [13 , 17 ] MF6 Solidity [13 , 17 ] MF7 Extent [13 , 17 ] MF8 Undulation characteristics [21 ] MF9 Compactness [20 , 29 ] MF10 Morphological Length of the ellipse major axis [20 ] MF11 Six morphological features (MF11–MF16) are extracted from the best-fit ellipse that approximates the size and position of the tumor Length of the ellipse minor axis [20 ] MF12 Ratio between the ellipse major and minor axes [20 ] MF13 Ratio of the ellipse perimeter and the tumor perimeter [20 ] MF14 Overlap between the ellipse and the tumor [20 ] MF15 Angle of the ellipse major axis [20 ] MF16 Morphological NRL entropy [18 , 20 ] MF17 Two morphological features (MF17-MF18) are extracted from the NRL of the tumor NRL variance [18 , 20 ] MF18