Existence of spicules around a mass is one of the important signs which characterize malignant tumors. In this study, we developed a new automated method for detecting spicules for each mass on mammograms and determined a spicule feature value. Digital mammograms were obtained with a laser film scanner with 0.1-mm sampling distance and 10-bit density resolution. After the smallest rectangular region containing the mass area was cut off and high-frequency components were enhanced by using an unsharp-mask filter, the magnitude components of density gradient vectors were calculated within the segmented mass area. The detection was performed by using a “multistage pendulum filter” which was newly developed. Then, the spicule feature value, which indicates the degree of “spicule presence” around the mass, was calculated from the detected spicules. The multistage pendulum filter was evaluated with 71 Japanese women's digitized mammograms, and it was able to show the most of the spicules that a radiologist pointed; the sensitivity of 88.5% and specificity of 80.0% were obtained, which demonstrates the effectiveness of our proposed method.
