UCSF

Magnetic resonance spectroscopic imaging (MRSI) is one of the most technically demanding areas of magnetic resonance imaging. To achieve the necessary performance, current methods employ many of the recent advances in RF pulse design. Extending MRSI methods to higher field strength poses a number of additional challenges, along with a wealth of opportunity, related to the design and use of RF pulses. With the availability of 3 T whole body clinical scanners, the potential exist for improving signal-to-noise ratio (SNR), spatial resolution and/or reduction of scan-time. Brain MRSI exams are a useful clinical tool for treatment planning and monitoring glial disease, but an overriding challenge has been the amount of coverage that could be achieved using this technique.

MR spectroscopic imaging was utilized to study the biochemical changes over time, and evaluate how these changes related to conventional imaging methods, in a population of low-grade glioma patients at 1.5 T. Although we were able to note unique metabolic changes in brain cancer tissue, the study ultimately was limited by SNR and coverage.

In order to potentially address the limitations in SNR observed at 1.5- T, MRSI was implemented at 3 T. The results demonstrated the acquisition of clinically acceptable spectra from brain tumor patients that were representative of the population encountered routinely in clinical practice. The improvements in sensitivity could be utilized either to detect more subtle differences in metabolite levels, or to reduce the spatial resolution.

Previously observed limitations in spatial coverage led to the design of the cosine-modulated very selective suppression (CM-VSS) pulse. This pulse design was not only optimized for improved coverage at 3 T, but it was also designed to allow more saturation bands in a shorter time period. With CM-VSS pulses, MRSI could now be feasibly acquired from almost all brain tissue.

The results of this dissertation demonstrate improvements in MRSI techniques that allow for better spatial characterization of brain tumors. It is expected that these improvements in data quality and tumor coverage will result in increased usage of MRSI sequences in clinical settings, and improved evaluation of brain tumor biochemistry.