Accurate and efficient quantitation of the data components in magnetic resonance spectroscopy (MRS) signals can be an important step in the analysis of biochemical substances. In several practical applications of MR spectroscopy the user is interested only in the data components lying in a small frequency band of the spectrum. A frequency-selective, or sub-band, analysis deals precisely with this kind of spectroscopy: estimation of only those spectroscopic components that lie in a selected frequency band of the (MR) data spectrum, with as little interference as possible from the out-of-band components and in a computationally efficient way. One obvious application for such a sub-band technique is to lower the computational burden in situations when the measured data sequence contains too many samples to be processed using a standard full-spectrum method. This thesis deals with several parametric methods to perform a frequency-selective data analysis. In addition, the possibility to incorporate prior knowledge about some of the components in the data is considered, a procedure that generally increases the parameter estimation performance significantly. A data model of exponentially damped sinusoids is assumed for the presented methods, which are applied to both simulated and experimental (in-vitro) MR data.
Keywords: frequency-selective spectroscopy; frequency-domain data processing; sub-band analysis; SVD-based parameter estimation; damped sinusoidal model.
Download BibTeX entry.