A Computational Approach Toward Identification of Malignant Lesions of the Human Breast: The Dynamics of Magnetic Resonance Imaging Contrast Agent Uptake
Abstract
Dynamic Gd-DTPA-2 contrast enhanced MRI images are currently used to help identify the malignancy of breast lesions. The time course of contrast agent uptake may thus be a non- invasive method for discriminating benign vs. malignant breast lesions. In this study, a computational technique called fuzzy clustering is employed to empirically categorize voxels of breast MRI images based on the time course of contrast agent enhanced image intensity. Several parameters of the clustering algorithm have been optimized including the fuzzy index and cluster membership thresholds. The clustering algorithm has been able to discriminate contrast uptake heterogeneity within breast lesions, and central vs. peripheral regions. These two features have not, on their own, discriminated malignant vs. benign lesions.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 1999
- Accession Number
- ADA385911
Entities
People
- John S. Leigh
- Jonathan H. Kaufman
Organizations
- University of Pennsylvania