In this paper, it is shown how the advantages of continuous requirements are met before applying a quantitation method. regularization (CR) can be exploited to achieve an improved, fully Here, the application of the Cadzow enhancement procedure automated LPSVD analysis of MRS time-domain data. The main (EP) (8) to MRS data leads to considerable improvements advantage of CR is its ability to determine the number of spectral (9). The EP algorithm restores the Hankel structure of the components even at low signal-to-noise ratios, which suggested its data matrix which results in a reduction of the noise in the use for in vivo spectroscopy. Estimation of the spectral parameters reconstructed signal. Recently, a new, improved enhanceis possible. Two alternatives of automated data-analysis schemes ment procedure (IEP) which minimizes the filtering effects are thoroughly investigated by means of Monte Carlo studies. The has been proposed (10).
results suggest the combination of CR for model-order estimation
All of the aforementioned SVD-based methods require with other methods for more-accurate parameter estimation. Several possible combinations, including those with an improved en-the number of signal components as an input from the user. hancement procedure and a total-least-squares method for quanti-To achieve fully automatic MRS quantitation, a method that tation, are discussed. Recommendations are given for spectral automatically and reliably determines the number of signal analysis, and a new data-analysis protocol which performs significomponents is needed. In order to be successful for in vivo cantly better than previously used protocols of the same type is MRS data, this method must be robust even at low signalproposed. α§ 1997 Academic Press to-noise ratios (SNRs). In this context, the application of the principle of continuous regularization (CR) (11) to the LPSVD analysis promises important beneficial effects. As