Local elastic matching and pattern recognition in MR mammography

Abstract

Contrast agent-enhanced magnetic resonance (MR) mammography is a highly sensitive method in breast cancer detection. The inflow of contrast agent is tracked to classify tissue as malignant or benign. In this article, we introduce approaches to overcome two of the most prominent problems in MR mammography: (a) variability in contrast agent dynamics, and (b) motion of the patient during data acquisition. The variability of contrast agent dynamics is considered using data-driven analysis based on self-organizing maps. Motion during data acquisition is handled by means of a fast local elastic matching algorithm. © 1999 John Wiley & Sons, Inc. Int J Imaging Syst Technol, 10, 199–206, 1999