14th IEEE Symposium on Computer-Based Medical Systems (CMBS'01)
Morphological Templates for Extracting Texture Information in X-Ray Mammography
Bethesda, Maryland
March 26-March 27
ISBN: 0-7695-1004-3
Abstract: This paper describes a new filter to enhance lesions in x-ray mammograms by extracting texture information within a morphological template. A circular ring template is used to characterize the morphological features for certain types of lesions. Three texture-related features, standard deviation, skew and kurtosis, are applied within the template for visual enhancement of the lesions. The results showed that the ring template enhanced visualization of a typical lesion by creating a "halo'' response surrounding the lesion. The standard deviation, skew and kurtosis were sensitive to roughness, presence of peaks and abrupt changes in the texture of the mammograms. Of the 45 lesions, 87% were enhanced using standard deviation, 84% by skew and 80% by kurtosis.