Detection of architectural distortion in prior mammograms using measures of angular dispersion

Architectural distortion is a mammographic sign of breast cancer at an early stage. We propose methods for the characterization and detection of architectural distortion based on measures of spicularity and angular dispersion of the patterns in automatically detected regions of interest (ROIs) using Gabor filters and phase-portrait analysis. The novel features for the characterization of the spiculating patterns of architectural distortion include an index of convergence of spicules computed from the Gabor magnitude and coherence using the Gabor angle response; radially weighted difference and angle-weighted difference measures of the intensity, Gabor magnitude, and Gabor angle response; and the angle-weighted difference in entropy of spicules computed from the intensity, Gabor magnitude, and Gabor angle response. At first, 4,224 ROIs were automatically obtained from 106 prior mammograms of 56 interval-cancer cases, including 301 true-positive ROIs, and from 52 mammograms of 13 normal cases using Gabor filters and phase-portrait analysis. Using the newly proposed features through feature selection and pattern classification, the best result achieved, in terms of the area under the receiver operating characteristic curve, is 0.76 with an artificial neural network based on radial basis functions. Free-response receiver operating characteristic analysis indicated a sensitivity of 0.90 at 6.3 false positives per patient.