Premium
An evaluation of new processing protocols for in vivo NMR spectroscopy
Author(s) -
Mazzeo A. R.,
Levy G. C.
Publication year - 1991
Publication title -
magnetic resonance in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.696
H-Index - 225
eISSN - 1522-2594
pISSN - 0740-3194
DOI - 10.1002/mrm.1910170219
Subject(s) - deconvolution , signal processing , computer science , data processing , multidimensional signal processing , singular value decomposition , fourier transform , convolution (computer science) , noise (video) , data set , algorithm , artificial intelligence , digital signal processing , mathematics , artificial neural network , mathematical analysis , computer hardware , image (mathematics) , operating system
In vivo NMR spectroscopy is often complicated with problems of low signal‐to‐noise, poor resolution, undefined peak shapes, and nonlinear baselines despite the efforts of investigators to optimize their experiments. Several data processing options are available to spectroscopists to enhance resolution and signal‐to‐noise and /or to flatten baselines. There is some question about how these processing protocols affect quantitative information. This paper evaluates five different processing protocols for their ability to extract quantitative information from a set of nonideal spectra. Three of the protocols involve recently developed statistical signal processing methods, maximum entropy Fourier spectral deconvolution, linear prediction singular value decomposition, and baseline deconvolution. These protocols are compared with the conventional processing methods of convolution difference and zeroing initial data points of the FID. The methods are evaluated by use of a quantitative 31 P model sample and also are demonstrated on surface coil 31 P data. © 1991 Academic Press, Inc.