Correction of proton resonance frequency shift MR-thermometry errors caused by heat-induced magnetic susceptibility changes during high intensity focused ultrasound ablations in tissues containing fat

Purpose In this study, we aim to demonstrate the sensitivity of proton resonance frequency shift (PRFS) ‐based thermometry to heat‐induced magnetic susceptibility changes and to present and evaluate a model‐based correction procedure. Theory and Methods To demonstrate the expected temperature effect, field disturbances during high intensity focused ultrasound sonications were monitored in breast fat samples with a three‐dimensional (3D) gradient echo sequence. To evaluate the correction procedure, the interface of tissue‐mimicking ethylene glycol gel and fat was sonicated. During sonication, the temperature was monitored with a 2D dual flip angle multi‐echo gradient echo sequence, allowing for PRFS‐based relative and referenced temperature measurements in the gel and T1‐based temperature measurements in fat. The PRFS‐based measurement in the gel was corrected by minimizing the discrepancy between the observed 2D temperature profile and the profile predicted by a 3D thermal model. Results The HIFU sonications of breast fat resulted in a magnetic field disturbance which completely disappeared after cooling. For the correction method, the 5th to 95th percentile interval of the PRFS‐thermometry error in the gel decreased from 3.8°C before correction to 2.0–2.3°C after correction. Conclusion This study has shown the effects of magnetic susceptibility changes induced by heating of breast fatty tissue samples. The resultant errors can be reduced by the use of a model‐based correction procedure. Magn Reson Med 72:1580–1589, 2014. © 2013 Wiley Periodicals, Inc.