A Proposed Methodology for Magnetic Resonance Images' Geometrical Distortion Correction Intended To Use For Radiotherapy Planning Using a Large Inhouse Field of View

Exquisite soft tissue contrast of magnetic resonance images (MRI) and the new combined radiotherapy system of MR-Linac have been the main impetus for applying MR imaging in radiotherapy. One limitation of MR-based radiotherapy is the geometric distortion of MR images that can generate errors in the contouring and dosimetry stages. This study aimed to evaluate and correct geometric distortion for radiotherapy applications.
A large field of view (FOV) phantom develop using Perspex sheets and 325 plastic pipes. The quantification and correction of MR images' system-related geometric distortion are conducted for HASTE protocol by MATLAB and 3D slicer software in phantom and patient images. The effect of MRI images geometrical distortion was evaluated for ten patients undergoing body radiotherapy treatment.  CT images were used as a primary dataset to estimate the distortion map.
The phantom investigation results indicate that in radial distances of < 13 cm (or FOVs < 25 cm), the amount of distortion is under 2 mm. Still, at more considerable radial distances, distortion may increase up to about 3.5 cm. MR images of Patients with lateral (LAT) and anterior-posterior (AP) diameters of more than 38 cm and 25 cm respectively, need to be corrected for geometric distortion.  
MR images' geometric precision in large FOVs is not sufficient for MRI only treatment planning of radiotherapy and further corrections are required. The B-spline deformable registration method can correct the MR geometric distortion until an acceptable range of 2 mm for radiotherapy applications.