Resolving nuclear magnetic resonance data of complex mixtures by three-way methods:: Examples of chemical solutions and the human brain

Despite the use of hyphenated and/or high-resolution instruments in analytical spectroscopy, the resulting spectral data often represent mixtures of several components. When no reference data in the form of reference spectra or concentration profiles are available, self-modeling mixture analysis techniques can be utilized to obtain the spectra of the pure components and their concentration profiles. There are many different algorithms to resolve mixture spectra, and the mathematical procedures involved are not always simple. This paper will discuss some of the aspects and problems of self-modeling mixture analysis, with the focus on the three-way method and without going into the mathematical details. Practical examples will be shown of methods applied to nuclear magnetic resonance data. The techniques discussed can also be applied to magnetic resonance images and an example will be shown of the human brain.