Magnetic resonance spectroscopic imaging (MRSI) shows great promise for basic physiological research and for clinical imaging of metabolic function. The use of MRSI to monitor metabolic alterations in myocardial ischemia, to localize and assess brain tumors and multiple sclerosis, to detect and evaluate breast lesions for malignancy, as well as to determine the seizure focus of temporal lobe epilepsy will all require increased spatial resolution while maintaining good spectral resolution. Increased resolution ordinarily requires more time to acquire data. Recent methods improve efficiency by adapting echo-planar imaging to MRSI, but these improvements are not sufficient in all cases. This project exploits prior information about the signal in the form of a time-domain model. The reduced degrees of freedom allow the same quality of reconstructed spectra with fewer temporal samples acquired in an echo-time-encoding experiment. However, given prior information, certain irregularly spaced sets of echo-time values provide significantly more information about the original spectrum than a regularly spaced set. This project will implement a temporal sampling acquisition method, exploit a known model of approximate relative spectral peak locations of the object being imaged to choose the temporal sampling scheme that minimizes the error in the reconstructed image, and reconstruct the image reliably and quickly from the acquired data. The project aims to: 1) Develop an algorithm that selects a sparse set of echo-time shifts which yield the most informative samples for the time-domain model parameters and allows reliable reconstructions from a subset of oscillating-gradient images acquired using sparse echo-time sampling. The optimal data points will then be acquired and fitted to these nonlinear time-domain models. 2) Validate the method on phantoms and real proton SI data. The result will be an MRSI acquisition & reconstruction technique that can provide images at a given resolution and SNR with only 10-40% of the acquisition time of existing techniques to make a wide variety of MRSI imaging applications feasible in practice. This research will impact public health by enabling a tool for physiological research as well as a diagnostic tool for obtaining images that are sharp enough to distinguish healthy tissue from damaged heart tissue, cancerous breast tissue, or brain tumors. This information will allow more accurate and precise treatment of diseased or damaged tissue. ? ? ?
