Comparison of Two Methods for Patient Specific Distortion Corrections for MR Only Treatment Planning
Weiss S, Nejad-Davarani S, Renisch S, and Glide-Hurst C. Comparison of two methods for patient specific distortion corrections for MR only treatment planning. Med Phys 2017; 44(6):3097-3098.
Purpose: While MRI-only treatment planning is becoming more widespread, a robust clinical solution for patient-specific distortion corrections is not currently available. This work explores an alternative approach for B0- mapping estimated from mDixon, often acquired for MR-only planning, compared to a dedicated dual-echo gradient-recalled echo (GRE) sequence, with the overarching goal of developing an efficient and robust approach for patient-specific distortion correction maps. Methods: Healthy volunteers were imaged in the pelvis and head and neck (H&N) regions at 1.5T and 3T. B0 field maps were generated with two approaches: (1) conventional: dualecho GRE (using two in-phase echo times) and (2) experimental: mDIXON using much shorter echo times. Additionally, the impact of acquisition resolution was evaluated for 1.7 9 1.7 9 5 mm (low resolution, LR) vs. 1.2 9 1.2 9 2.4 mm (high resolution, HR) in H&N. Distortion maps were generated from B0 field maps based on bandwidth and acquisition pixel size and compared between approaches. Results: In H&N, mDIXON revealed similar distortion magnitudes between field strengths (<0.5 mm). Distortions were higher near ear canals (∼1 mm), sinuses (∼1 mm) and dental fillings (>1 mm). HR B0 maps were more sensitive at interfaces than LR, although LR was suitable given its overall accuracy in bulk voxels and shorter acquisition time (40% of HR). LR conventional GRE was well-approximated by mDIXON: 96% (1.5T) and 99% (3T) of voxels estimated by mDixon differed from their GRE maps by <0.5 mm. In the pelvis, 99% of GRE distortion voxels and 100% of mDixon distortion voxels were within ±0.3 mm. Differences between techniques occurred near regions with high spatial variation and with phase unwrapping errors. Conclusion: mDIXON closely approximated GRE for patient-specific distortion assessment. Slight differences observed near tissue interfaces require further assessment to determine geometric and dosimetric impact. mDIXON-derived B0 maps may be advantageous for integration into in-line processing without requiring additional sequences.