Segmentation of Fibroglandular Tissue for Measuring Volumetric Breast Density




          We have developed a comprehensive segmentation method using a fully automatic program to segment the fibroglandular tissue on 3D MR images. The procedures include: breast segmentation using thoracic template, chest wall muscle excluding based on b-spline fitting, skin exclusion based on dynamic search, bias-field correction using N3+FCM algorithm, and lastly separation of fibroglandular and fatty tissue using fuzzy-c-means. In addition to the total volume, we can also measure the morphological parameters based on circularity, convexity, and irregularity analysis.