Signal Processing for Highly Resolved 2D NMR

Abstract

As the name implies, the “traditional” processing scheme for Fourier transform nuclear magnetic resonance (FT-NMR) signals is centred on the Fourier transform. However, other techniques can either replace or supplement the Fourier transform: extracting more information from fewer datapoints, improving sensitivity and/or resolution, reducing acquisition time (while maintaining spectral quality) and even reconstructing spectra whose experimental acquisition is too time-consuming to be feasible. Following an overview of “traditional” FT-NMR processing, including an analysis of apodization functions, this chapter will discuss alternatives to the Fourier transform applicable to 2D spectroscopy, including compressed sensing and covariance NMR. This chapter will evaluate processing techniques in light of the specific advantages of 2D NMR, such as the inherent ability to treat 2D datasets as matrices subject to well-studied matrix operations as well as the symmetry of certain 2D NMR experiments. On the other hand, this chapter will address certain challenges in processing rapidly acquired 2D NMR spectra, such as crowded signals and the inapplicability of certain multidimensional processing techniques to data with only a single indirect dimension. This chapter will also review software for NMR signal processing, such as NMRPipe and Mnova.