Time-domain analysis of magnetic resonance spectra and chemical shift images

The utility of adaptive prediction and filtering algorithms and the autocorrelation-based Yule-Walker algorithm to predict and filter complex NMR (nuclear magnetic resonance) data is demonstrated. The application of these methods improves the available signal-to-noise ratio using time-domain analysis, and increases the low resolution via prediction algorithms in data containing phase errors introduced by hardware limitations. The application of the complex least-mean-squares and the modified-least-mean-squares transversal and lattice algorithms to low- and high-resolution NMR data records is demonstrated. The resolution and windowing problems found in the discrete Fourier transform are overcome by these alternative methods