Introduction: Gadolinium enhanced MRI has been widely recognized as an extremely sensitive test for breast cancer, capable of detecting a significant number of breast cancers which are obscured on mammograms, or not-palpable. Breast MRI has been plagued by non-specificity. Not all lesions which enhance are cancer. We have pursued two simultaneous methods to improve breast MR specificity: characterization of suspicious lesions based on their morphology on high resolution 3DSSMT images, and assessment of lesion neo-vascularity based on a 2-compartment pharmacokinetic model of rapid dynamic spiral images Methods: High resolution, 3DSSMT imaging was developed at Stanford several years with over 100 exams have been performed. We retrospectively performed a systematic review of all 3DSSMT exams, and evaluated all pathologically proven lesions on the basis of their morphology, including lesion borders, internal architecture, and patterns of enhancement. ROC analysis was performed to determine which characteristics were most predictive of malignancy. Ultrafast whole breast dynamic imaging with Spiral MR was developed and pioneered at Stanford about 2 years ago. A 2-compartment pharmacokinetic model was implemented to analyze these lesions, and yielded a number of parameters, including the vascular permeability constant K21, the extraction constant Kel, the normalized amplitude, and secondary parameters such as the wash-in rate, and wash-out rate. ROC analysis was used to assess the diagnostic performance of the parameters, and compare their ability to separate benign and malignant disease. Results: High resolution 3DSSMT, and rapid dynamic Spiral imaging were found to improve specificity of contrast enhanced breast MRI. Conclusions: New advances in breast MR, promise to increase the specificity of the technique. Analysis of lesion morphology reveals that some features, such as smooth borders, internal septations, and completely uniform enhancement suggest benign disease, whereas spiculated, ring enhancement, and focal skin thickening are usually associated with malignancy. Pharmaco-kinetic modeling suggests that the K21 parameter best distinguishes cancer from benign disease. Three-dimensional rendering of breast MR images aided surgical management in a majority of a limited series of cases.
Showing the most recent 10 out of 446 publications
