Purpose:To improve the immunity of the proton resonance frequency shift (PRFS) method of MRI temperature mapping against magnetic field disturbances. Since PRFS is a phase-sensitive method, it misinterprets magnetic field disturbances as artifact temperature changes. If not corrected, the resulting temperature artifacts can completely obscure the true temperature estimation, especially if the temperature elevations are small.
Materials and Methods:Since the fat protons experience the same magnetic field disturbances as the water protons, but no temperature-related frequency shift, the fat signal has been used for correcting PRFS temperature maps for the disturbances. A simple correction method is proposed that has either better compensation capability than the phase correction methods previously reported or higher spatial and temporal resolution than the spectroscopic correction methods previously reported. The evaluated method is based on the utilization of several gradient and spin echoes acquired within one repetition interval with waterand fat-selective scans.
Results:In a series of phantom experiments, the improved method is shown to enable the reconstruction of accurate temperature maps in spite of interscan motion, suboptimal fat-water separation, and a wide range of magnetic field disturbances.
Conclusion:Our approach can be used for the guidance of thermal therapies involving tissues containing fat or surrounded by fat. THE ABILITY TO MEASURE small temperature changes, with high accuracy, using MRI, has many potential clinical applications; for instance, specific absorption rate (SAR) management or thermal ablation surgery guidance. The proton resonance frequency shift (PRFS) method is the most widely used method for MRI temperature monitoring because it does not depend on the imaged tissue and thus does not require any calibrations for a particular patient or anatomy. However, the temperature sensitivity of the method comes from the phase accumulated by water protons before the echo acquisition. In addition to the temperature-related term, the accumulated phase also contains contributions from all factors influencing the local magnetic field within the imaged anatomy; for instance, main and shimming field drifts, susceptibility changes resulting from breathing and/or anatomy repositioning, tissue property and perfusion changes with temperature, etc. As the demand for the increase of spatial, temporal, and temperature resolution of MRI temperature maps is growing, these disadvantages become even more restrictive.Since the fat signal does not exhibit the temperaturedependent resonant frequency shift, the phase maps and spectra acquired during fat-selective acquisitions have been utilized for correcting the deleterious effect of field disturbances onto the resulting PRFS temperature maps (1-3). However, magnetic field disturbances are known to influence the quality of fat-water separation, making fat-and water-selective pulses less selective and adding undesired signal from the "to-be-suppressed" componen...