Frequency-selective Quantiication of Biomedical Magnetic Resonance Spectroscopy Data 1 Frequency-selective Quantiication of Biomedical Magnetic Resonance Spectroscopy Data

In this paper we focus on model tting techniques in both the time and frequency domain for analysis of biomedical magnetic resonance spectroscopy (MRS) signals. We examine the possibility to obtain accurate estimates of the parameters of selected peaks in the presence of unknown or uninteresting spectral features. We denote this by frequency-selective parameter estimation. A number of existing approaches are revisited and a new time-domain technique based on minimum-phase nite impulse response (FIR) lters is presented. The proposed method is compared to the application of a weighting function in the time domain, to frequency domain tting using a polynomial baseline and to the time-domain HSVD lter method. The HSVD lter method and the FIR method clearly outperform the other methods. The ease of use and low computational complexity of the FIR lter method make this an attractive approach for frequency-selective parameter estimation. The results are illustrated using relevant 13C and 31P MRS examples.