Derivation of quantitative information in NMR imaging: a phantom study.

The use of NMR imaging as a quantitative research tool requires insight into the relationship between various imaging techniques and their resultant images. Work was undertaken to elucidate this relationship by using the following procedure. First, a theoretical model of NMR imaging under various pulse sequences was elaborated. Subsequently, a series of inversion recovery and saturation recovery images of a particular object slice was generated by varying the sequence parameters. Finally, pure rho, T1 and T2 images of that slice were obtained by solving the corresponding model equations. This procedure was applied to a test phantom containing tubes with suitable reference substances, including aqueous solutions of agar, manganese chloride and deuterium, and water-fat mixtures. The concentration of various samples was chosen such as to yield rho, T1 and T2 values usually encountered in clinical NMR imaging. Experiments were carried out with a prototype resistive NMR imager with a static magnetic field of 0.14 T, corresponding to a hydrogen proton resonance frequency of 5.9 MHz. For most samples a weighted non-linear regression analysis showed the theoretical model to produce an adequate parametrisation of the data at the 5% significance level, given the number of data points and the experimental accuracy. The quantitative information extracted from the NMR imaging experiments, i.e. rho, T1 and T2, appeared to be in good agreement with the results of conventional methods, including NMR spectroscopy. The clinical efficacy of the proposed methods is currently being investigated.