The overall goal of this application is to develop totally automated spectral analysis and data processing methods for image formation of calibrated metabolite distributions in the brain, obtained using proton Magnetic Resonance Spectroscopic Imaging (MRSI) with short TE acquisition. The expected outcome is improved accuracy, standardization, simplified implementation, and improved image quality, which will expand the diagnostic applications of this imaging modality. The methods will be applied to ongoing clinical studies in brain, including: normal subjects; metabolic effects of aging; Alzheimer's disease; epilepsy and other brain diseases. To achieve this goal, the development of advanced, though computationally intensive, spectral analysis methods is proposed, to provide totally automated formation of metabolite images. Parametric methods will be implemented, using the maximum possible a priori information together with a semi-parametric approach to account for uncharacterized baseline signals. Inhomogeneously broadened line shape functions will be determined prior to spectral analysis to improve the parametric model, and additional processing will be used to account for common instrumental related distortions and limitations including large residual water and lipid signals. A detailed 1H spectral database will contain a description of all multiplet resonances with phase values to account for J-modulation effects with each acquisition sequence used. To determine this data for the 'low field' situation where strong coupling effects and localization pulse sequences complicate the observed MR spectral characteristics, a computer simulation program will be used and results verified using experimental studies. Following image formation, calibration procedures will be developed to provide metabolite images in a reproducible scale, enabling between subject comparison of all metabolite concentrations, either obtained at different locations, times or locations. Accuracy of spectral analysis and reproducibility of the measurement will be verified using computer simulation and experimental measurements in phantoms.
Showing the most recent 10 out of 23 publications
