MAXIMUM-LIKELIHOOD DECONVOLUTION OF NMR-SPECTRA IN MULTIDIMENSIONAL SPACE

Maximum-likelihood spectral deconvolution can be applied to three- and four-dimensional NMR spectra, improving resolution by effectively sha rpening lines in air of the dimensions of the data set. Spectral noise is simultaneously suppressed and cross-talk between (hyper)planes is minimized. This method is demonstrated for two three-dimensional data sets; the method can be combined with linear prediction time extension of multidimensional free-induction decays and other preparatory data processing. Multidimensional MLM reconstruction offers promise for fut ure fully automated data processing and analysis of large NMR data arr ays used far determination of complex molecular structures from NMR. T he algorithm is readily adapted to parallel computers, which will exte nd its practicality for the largest data arrays, including four- and f ive-dimensional data. (C) 1994 Academic Press, Inc.