Evaluation of a 3D Stack-of-Stars MR Sequence for Respiratory Hysteresis Characterization In a Motion Phantom
Abstract
Purpose
To evaluate the motion measurement accuracy of transverse and coronal 3D stack-of-stars MR sequences relative to 4D-CT and physical ground truth, using a phantom undergoing hysteretic and non-hysteretic 3D sinusoidal motion.
Methods
A Quasar MRI⁴ᴰ Motion Phantom (Modus Medical Devices) with a 30mm spherical insert filled with diluted gadolinium contrast was programmed to produce sinusoidal respiratory motion with known ground truth displacements (with hysteresis: right-left (RL):1.99mm, anterior-posterior (AP):15mm, superior-inferior (SI):40mm; without hysteresis: RL:1.99mm, AP:6.75mm, SI:39.8mm). The phantom was imaged on a 1.5T Siemens Magnetom Sola MR scanner in transverse and coronal orientations, and on a CT simulator (GE Discovery RT590), using 10-bin respiratory sorting in both modalities. Displacements in SI, AP, and RL directions were measured using centroid motion from intensity-based contouring of the sphere in the maximum bin for each direction of motion. Mean displacements for each axis and hysteresis condition were compared against the ground truth and the gold standard 4D-CT measurement to assess agreement within a pre-defined per-axis tolerance derived from typical abdominal motion and planning margins. A Welch’s ANOVA test followed by a Games-Howell post-hoc test were used to test statistical differences between each scan type at three significance levels: p<0.05, p<0.01, and p<0.001.
Results
4D-CT demonstrated RL estimates 2.00±0.02mm from the ground truth. In all other directions, there were no significant differences between any scan type, and all values were within 1.5mm of each other and 3mm of ground truth. Hysteresis does not affect the ability of MR to accurately measure displacement in any direction. The CT images exhibited reduced visibility of the sphere compared to the MR images, which increased ambiguity in target localization.
Conclusion
Measurement similarity combined with improved soft-tissue contrast improving target delineation supports 4D-MR as a potential suitable substitute for 4D-CT for abdominal patients undergoing MR-only adaptive radiotherapy.