Magnetic Resonance imaging (MRI) has potential to greatly improve detection and staging of breast cancer. The major problem which must be addressed before MRI can be used in this capacity is its lack of specificity. MRI does not provide enough information regarding lesion composition, shape, and contrast media uptake to reliably differentiate between benign and malignant lesions. This is due in part to strong magnetic susceptibility gradients in tumors and heterogeneous lipid signals which cause blurring, loss of contrast, poor edge delineation, and distortion - and thus reduce the ability of radiologists to characterize lesions. These artifacts can be minimized if high resolution proton spectra are acquired for each image voxel. We believe that this can be accomplished through use of Fast Spectroscopic Imaging (FSI) to calculate what the proton resonance in each voxel would look like in the absence of magnetic susceptibility and chemical shift effects. We have used FSI to obtain excellent separation between water and fat signals and significantly improved delineation of edges and texture with clinically acceptable run times in animal models. The goal of the work proposed here is to implement FSI pulse sequences and data analysis on a whole body scanner, optimize FSI using a phantom, test FSI with and without contrast in patients with suspicious breast lesions, and compare FSI with conventional MRI, mammography, and biopsy results. Successful completion of the proposed work will produce FSI methods which can easily be used in a clinical setting. This would be a significant advance in clinical MRI. Specifically - FSI with high spectral and spatial resolution would increase image contrast and signal-to-noise ratio, and improve edge delineation, image resolution and the accuracy of positional information.
Showing the most recent 10 out of 30 publications
