LAUSR

PURPOSE To improve the temperature monitoring precision using multi-echo proton resonance frequency shift-based thermometry with view-sharing acceleration for MR-guided laser interstitial thermal therapy (MRgLITT) on a 0.5T low-field MR system. METHODS Both precision and speed of the temperature measurement for clinical MRgLITT treatments suffer at low-field, due to reduced image signal-to-noise ratio (SNR), decreased temperature-induced phase changes, and limited RF receiver channels. In this work, a bi-polar multi-echo GRE sequence with a temperature-to-noise-ratio optimal weighted echo combination is applied to improve the temperature precision. A view-sharing-based approach is utilized to accelerate signal acquisitions while preserving image SNRs. The method was evaluated using ex-vivo (pork and pig brain) LITT heating experiments and in-vivo (human brain) non-heating experiments on a high-performance 0.5T scanner. RESULTS 1) After echo combination, multi-echo thermometry (i.e., 7.5 ms ~ 40.5 ms, 7 TEs) provides around 1.5 ~ 1.9 times higher temperature precision than the no echo combination case (i.e., TE7 = 40.5 ms) within the same readout bandwidth. Additionally, echo registration is necessary for the bi-polar multi-echo sequence. 2) For a threefold acceleration, the view-sharing approach with variable-density subsampling shows around 1.8 times lower temperature errors than the GRAPPA method. Particularly for view-sharing, the variable-density subsampling performs better than the interleave subsampling. 3) Ex-vivo heating and in-vivo non-heating experiments demonstrated that the temperature accuracy was less than 0.5 ° C and the temperature precision was less than 0.6 ° C using the proposed 0.5T thermometry. CONCLUSIONS View-sharing accelerated multi-echo thermometry is a practical temperature measurement approach for MRgLITT at 0.5 T.