Over the past year we have recruited approximately 150 patients into this project. We have done a detailed analysis of the ability of architectural features to distinguish benign from malignant lesions. Interpretation model using architectural characteristics extracted from breast MRI was shown to have a 96% sensitivity and 79% specificity. The 97% negative predictive value may make this technique valuable to help patients with suspicious mammograms avoid biopsy in many cases. To date, the model only incorporates architectural features from breast MR images. However, our rapid 3-dimensional technique allows us to also extract data regarding enhancement kinetics. Work is ongoing to incorporate enhancement kinetics into our interpretation model. Utilizing the 4 Tesla facility at the University of Pennsylvania we have been able to acquire images with 150 micron in plane resolution and 700 micron slice thickness. It is expected that this will be further improved with the gradient upgrade now being performed on the 4 Tesla system. However, the ultra high resolution images clearly demonstrate high architectural detail than our normal 300 micron images. Work establishing the diagnostic utility of the enhanced resolution is ongoing.
Showing the most recent 10 out of 414 publications
