A Phantom-Based Method to Measure MRI Distortion from Pedicle Screws for Spine Radiotherapy Planning
Abstract
Purpose
Magnetic resonance imaging (MRI) is an important tool for visualizing tumor sites and organs at risk during spine radiation treatment planning. Some people with spinal metastasis require pedicle screw fixation to stabilize the spinal column which causes metal-related artifacts and geometric distortions that limit MRI usability for radiation treatment planning. While metal artifact reduction techniques exist, tools comparing their performance are limited. Our objective was to develop a phantom and analysis method to quantify MRI geometric distortion around pedicle screws.
Methods
A phantom containing 84 reference markers and 2 pedicle screws was constructed. MRI scans of the phantom were collected (T2-weighted 2D turbo spin echo, TR/TE = 7980/83-87 ms, resolution = 0.938 x 0.938 x 3 mm3) with screws present, and once with the screws removed. The screw-absent MRI data was rigidly registered to one screw-present MRI scan with additional slices to include registration markers. Subsequent scans with screws only included phantom region, and data with three pixel bandwidths (100, 399, 698 Hz) were collected and compared. Reference markers were automatically detected in predefined slices, with manual correction as needed. Geometric distortion was quantified as the Euclidean distance between reference marker centers on MRI scans with/without screws.
Results
Six reference markers were excluded due to inconsistent detection across acquisitions. Automatic identification was successful for 82%, 82%, and 68% of reference markers at 698, 399, and 100 Hz bandwidth, respectively. Average distortion across reference markers per slice was 0.3–1.4 mm (698 Hz), 0.5–1.5 mm (399 Hz), and 0.6–5.5 mm (100 Hz). MRI distortion was generally highest at the lowest bandwidth, as expected.
Conclusion
Our phantom and analysis pipeline provides an approach for quantifying bandwidth-dependent MRI geometric distortion near pedicle screws. Future work will apply this framework to evaluate more advanced metal artifacts reduction techniques.