Purpose: MRI is widely applied in Stereotactic Radiosurgery (SRS) planning. Image fusion and distortion correction algorithms are deemed essential parts of the process. QA guidelines should be followed to ensure correct patient treatment (TG132). This work aims to develop tools and methods for geometric distortion assessment, to verify the corrections performed by Elements (Brainlab) TPS by using a virtual anthropomorphic phantom
Methods: Images from virtual, anthropomorphic, cranial phantom Brainweb were used (MRIo). From it, a synthetic CT with identical geometry was generated. Known distortions were introduced to the MRIo study to produce a deformed dataset (MRId). MRId was fused to the CT and distortion correction was applied by Elements’ Distortion Correction Cranial application. A set of corrected images (MRIc) was generated. Quantitative comparisons were performed between MRIo and MRId, andMRIo and MRIc, using two locally developed software tools. The first (Matlab), uses a 3D virtual markers arrangement overlaid on the MRI and calculates their displacements. The second (Python) makes a voxel-to-voxel comparison with a DTA-based metric
Results: Both tools evaluated the distortions applied to MRIo and the residual remaining distortion after the correction process. Mean deviation between the markers in MRIo and MRId was 1.9mm, with a maximum of 4.2mm. Mean residual deviation between MRIo and MRIc was 0.2mm with a maximum of 2.3mm. DTA analysis showed 77.7% of the voxels had displacements lower than 2mm in MRId with respect to MRIo, and 90.8% lower than 3mm. In MRIc, 98.1% of the voxels presented displacements lower than 2mm with respect to MRIo, and 99.8% lower than 3mm. DTA tool also verified that the presence of the markers did not alter the distortion correction process by Elements TPS
Conclusion: Software tools developed and virtual anthropomorphic phantom implementation allowed for robust quantitative verification of Elements’ distortion correction process
Not Applicable / None Entered.