A NOVEL 1-D BLOCK PROCESSING APPROACH TO 2-D NMR SPECTROSCOPY

In order to extract diagnostic information from NMR data, an imaging technique based on two-dimensional (2-D) spectral estimation has been previously developed. In this method, 2-D state realization is employed to identify protein frequencies using two open-loop matrices derived from row and column directions of the data set. The frequencies along the two directions are paired by diagonalization of one of the matrices using eigenvectors of the other. Computation of the open-loop matrices and their coupling, resulting from the diagonalization, increases complexity and renders realtime processing difficult. In this paper, we present a 1-D state realization algorithm that enhances the computational efficiency by using state matrices derived from only the row direction of the data. Thus, matrix coupling, pole pairing and computation of molecule amplitudes, required to model NMR data in 2-D state realization, are avoided. Resulting images from 1-D algorithm exhibit spectral resolution comparable to 2-D, and provide accurate estimates of the protein frequencies