The ability to map the spatial variation of the absolute, rather than the relative value of the equilibrium magnetization could be advantageous in many areas of NMR. However, direct measurement of M 0 is usually difficult because of the multiparametric dependence of the NMR signal. Here we propose a technique for mapping the spatial variation of the absolute value of M 0 , independent of relaxation weighting and flip angle calibration. This method, which works best at high field strengths, is based on the effect of the dipolar field due to the nuclear magnetization that is normally neglected in liquid-state NMR. The experimental implementation of this sequence at 3.0 T is described, and its initial application to the measurement of the water content of brain tissue is outlined. Mapping the spatial variation of the equilibrium nuclear magnetization, M 0 , can provide important information, which in the context of medical imaging can be used in treatment planning for neutron therapy (1) and microwave hyperthermia (2), as well as in evaluating the tissue : blood partition coefficient, -a parameter needed in the calculation of regional perfusion using arterial spin-tagging techniques (3). However, absolute measurement of M 0 is difficult in NMR experiments for several reasons. First, the multiparametric nature of the NMR signal means that its magnitude depends in a complicated manner on a combination of the pulse sequence parameters, including echo time (TE), repetition time (TR), and radiofrequency (RF) flip angle, and the NMR relaxation times T 1 and T 2 . Therefore, calculation of M 0 generally requires accurate knowledge of the relaxation-time weighting and flip-angle dependence, whose measurement may require multiple experiments to be performed (2-5). In imaging the distribution of M 0 , the spatial variation of all of these parameters must also be evaluated. A further difficulty relates to the often unknown conversion factor, K, relating the strength of the detected NMR signal to the amount of precessing magnetization. This factor depends upon the frequency of precession, sensitivity of the RF coil, and the gain of the spectrometer. Measurement of K usually involves comparing the signal of interest with that from a reference sample or compound of known concentration. The relaxation-time weighting and flip-angle dependence of the signal from such a sample will not necessarily be the same as that of the signal of interest, particularly in the case of an imaging experiment in which an external reference sample is employed. Therefore, these parameters should also be measured for the reference sample to allow proper calculation of M 0 .In recent imaging experiments, for example, Roberts et al. (3) measured the spatial variation of the equilibrium magnetization in the human brain by fitting to data from a progressive saturation experiment. They corrected for the effects of RF inhomogeneity by using similar data acquired from a spatially uniform phantom of known water content, and also used a reference sample of anti...
