Abstract

Two new approaches are described for speeding up the acquisition of multidimensional NMR spectra. The first method uses sparse radial sampling of the time-domain data instead of the systematic recording of all the elements of the multidimensional array. Fourier transformation of skew sections through the time-domain signal generates plane projections of the frequency-domain spectrum. From a small number of such projections the full multidimensional spectrum can be reconstructed. The second technique economises measurement time by restricting the excitation to selected chemical sites, using prior information about the NMR frequencies from a conventional spectrum. This involves an encoding and decoding scheme based on Hadamard matrices. For isotopically-enriched samples, this Hadamard method can be exploited to highlight the important sites, as if the molecule had been specifically enriched at the chosen locations.

Keywords: Hadamard, Multidimensional, NMR, Projection-reconstruction, Proteins