Technical Optimization of SyntheticMR for the Head and Neck on a 3T MR-Simulator and 1.5T MR-Linac: A Prospective...

The purpose of this study was to optimize the technical tradeoffs associated with integrating the quantitative maps available from SyntheticMR into the head and neck adaptive radiation oncology workflow. Recent work has begun to investigate SyntheticMR in the adaptive radiation oncology workflow, however no studies have investigated the variation in acquisition parameters and their relationship to the resulting quantitative maps. Filling this gap will facilitate SyntheticMR's translation to the adaptive radiation therapy setting in the head and neck due to the reduced bias and increase repeatability and reproducibility of its generated quantitative relaxometric biomarkers.

The 2D multi-echo multi-delay (MDME) sequence offered from SyntheticMR was acquired on a MR-Simulation and MR-Linac device in both a phantom and healthy volunteers. Scans were optimized across acceleration factors, slice gaps, acquired voxel sizes, repetition times, echo times, refocusing flip angles, noise-limited gradients, number of slices, and echo train lengths. Quantitative relaxometric T1, T2, and PD maps were tested for accuracy using both correlation and mean absolute bias analysis. Noise profiles were evaluated using the coefficient of variation (CoV) in uniform regions of interest.

The following main findings were reported: (1) noise-limited gradients did not affect the bias or CoV, (2) increasing acceleration factor did not affect bias, but it significantly increased CoV, (3) increasing the number of slices resulted in different significant changes in the quantitative parameters between the MR-Simulator and MR-Linac, (4) reducing the echo train length led to generally reduced bias and CoV, and (5) the repeatability CoV was within previously reported literature values on similar scanners.

The behavior of the 2D-MDME sequence from SyntheticMR was characterized across a wide range of acquisition parameters designed for ideal head and neck imaging on both the MR-Simulation and MR-Linac devices in phantom and healthy volunteers. Significant differences existed across several acquisition parameters which should be accounted for when making adjustments for application-specific field-of-views, scan time, and desired ranges of T1, T2, and PD. Application of these findings to the development of efficient head and neck adaptive radiation therapy MRI protocols can enrich the current quantitative biomarker landscape allowing for more informed treatment adaptations.