Automatic Labelling and BI-RADS Characterisation of Mammogram Densities

• Published in: 2005 IEEE Engineering in Medicine and Biology 27th Annual Conference Vol. 2005; pp. 6394 - 6398
• Main Authors: Marias, K., Linguraru, M.G., Ballester, M.A.G., Petroudi, S., Tsiknakis, M., Brady, M.
• Format: Conference Proceeding Journal Article
• Language: English
• Published: United States IEEE 2005
• Subjects: Adaptive filters; Anisotropic magnetoresistance; Biomedical imaging; Breast; Image Analysis and Classification; Image edge detection; Image segmentation; Intelligent Decision Support Systems; Labeling; Laboratories; Muscles; Shape
• ISBN: 0780387414; 9780780387416
• ISSN: 1094-687X; 1557-170X; 1558-4615
• DOI: 10.1109/IEMBS.2005.1615961

Intelligent management of medical data is an important field of research in clinical information and decision support systems. Such systems are finding increasing use in the management of patients known to have, or suspected of having, breast cancer. Different types of breast-tissue patterns convey semantic information which is reported by the radiologist when reading mammograms. In this paper, a novel method is presented for the automatic labelling and characterisation of mammographic densities. The presented method is first concerned with the identification of the prominent structures in each mammogram. Subsequently, 'dense tissue' is labelled in a mammogram data set, and BI-RADS classification is performed based on a 2D pdf that is contracted from a "ground truth" data set as well as a shape analysis framework. The presented method can be used in large-scale epidemiological studies which involve mammographic measurements of tissue-pattern, especially since breast-tissue density has been linked to an increased risk of breast cancer