20 January 1977 Texture Analysis In Xeromammograms
Author Affiliations +
A relatively new x-ray imaging technique, xerography, gives very detailed, high contrast images. It is particularly useful in detecting texture changes associated with early breast cancer. We describe a process of digitally scanning photographs of these xeromammograms. Normalization of the data representing the breast images is necessary to later classification; four methods were used for our study. The visual appearance of certain abnormal texture is described and used to motivate our statistical measures of texture. One group of measurements is derived from density value histograms; a second group depends on a density value co-occurrence matrix; a third group computes statistics from a gradient of the image; the last group analyzes the number of connected paths in the image. Nineteen features are collected from these measurements and are evaluated by using them to classify abnormal from normal tissue in a set of four test cases.
© (1977) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Carolyn Kimme, Carolyn Kimme, } "Texture Analysis In Xeromammograms", Proc. SPIE 0089, Applications of Optics in Medicine and Biology, (20 January 1977); doi: 10.1117/12.955039; https://doi.org/10.1117/12.955039


Back to Top