Purpose: Single-reference variable flip angle Stack-of-Stars (SR-VFA-SoS) thermometry with KWIC reconstruction provides simultaneous aqueous and fatty tissue temperature monitoring. This work evaluates the effect of k-space weighted image contrast (KWIC) reconstruction of a SR-VFA-SoS sequence on the accuracy of both proton resonance frequency shift (PRFS) and T 1 -based thermometry with different temporal footprints, spatial resolutions, echo times, and heating durations. Methods: Magnetic resonance-guided focused ultrasound (MRgFUS) sonications were performed using an asymmetric KWIC-reconstructed single-reference variable flip angle Stack-of-Stars (SR-VFA-SoS) sequence and a validated segmented echo planar imaging (segEPI) thermometry sequence in tissue-mimicking phantoms and an in vivo rabbit model. KWIC windows of projection length 89–377, voxel resolutions of 1.2 mm isotropic and 1.5 × 1.5 × 2.0 mm, three different sets of echo times, and heating durations of 30.2–123.3 s were evaluated. Peak temperature accuracy, calculated cumulative thermal dose (CTD), and temporal lag were assessed. Results: In phantoms and the animal model, significant differences in peak temperatures were observed between the SR-VFA-SoS and segEPI thermometry at 1.5 × 1.5 × 2.0 mm voxel size, but not 1.2-mm isotropic. Temperature discrepancies were independent of echo time. Prolonged heating duration with lower power at the larger voxel size resulted in lower peak temperature error. Asymmetric KWIC-reconstructed SR-VFA-SoS peak temperatures were consistently temporally shifted later than segEPI peak temperatures. Conclusion: SR-VFA-SoS thermometry with KWIC reconstruction offers promising temporal resolution for simultaneous aqueous and fatty tissue temperature monitoring during MRgFUS. However, temporal blurring due to KWIC reconstruction results in underestimation and temporal shifting of peak temperatures for large voxel sizes and short heating durations. These observations highlight the need for careful KWIC and MRgFUS parameter optimization.
Richards et al. (Wed,) studied this question.