Automated Region of Interest Detection of Spiculated Masses on Digital Mammograms

摘要

We have developed a novel technique to automatically identify a region of interest (ROI) surrounding a spiculated lesion on a mammogram. Our proposed approach for determining the size of the ROI depends on the response of a set of unique Spiculation Filters (SF). The design of these filters is based on manually annotated physical characteristics of spicules. The accuracy of our algorithm is measured in terms of the percentage of spicule pixels located inside the identified ROI. Spicules on each image were identified by an experienced radiologist to serve as a reference to determine the percentage of spicules located in the ROI. On average, 94 percent of spicule pixels were located inside the ROI identified by our algorithm.