Magnetic resonance mammography (MRM) is an imaging technique under development to overcome the limitations of X-ray mammography. One method uses permeability-surface area (PS) products derived from dynamic contrast enhancement (DCE) data to improve specificity. The specificity, 37 to 97%, obtained by this technique is controversial. DCE requires high temporal resolution to collect the kinetic information at the expense of spatial resolution. No one has reported if the averaging of the PS product, induced by partial volume effects, limits the effectiveness in using PS products to differentiate and grade tumors. We are testing the hypothesis that the spatial resolution used in human DCE MRM results in partial volume averaging that significantly reduces the prognostic information obtainable with PS products calculated from DCE-MRM data, and that reduced encoding methods that reconstruct images based on one reference image produce PS-products with partial volume averaging. We then demonstrate a technique to obtain images with both high temporal and spatial resolutions. We will accomplish these goals by studying the PS product of Gd(III)-DTPA in tumor region of interests in N-ethyI-N-nitrosourea induced rat mammary tumors as a function of both the in plane resolution and slice thickness. The PS values are calculated with a two compartment model at in plane resolutions that include those used clinically, 6.25 mm2, and at the """"""""microscopic field"""""""" sizes used to analyze vascular density, 0.74 and 0.152 mm2, in vitro. PS values are correlated to tumor grade, vascular density, and vascular permeability factor obtained by in vitro histochemical methods. We apply a reduced encoding method to obtain DCE MRM data with both high temporal and spatial resolutions, and show that the PS product calculated from these images are accurate relative to those obtained with standard high spatial resolution techniques. The algorithm uses a generalized series method with both pre and post contrast enhanced reference images, TRIGR. The dynamic data obtained with low in plane resolution is reconstructed to high spatial resolution with TRIGR. This maintains the high temporal resolution. The protocol is then used with DCE MRM and a dendrimer based contrast agent to differentiate benign from malignant tumors and compared against histological methods.
Showing the most recent 10 out of 15 publications
