Click here to

Session: CT and MR-Guided Adaptive RT [Return to Session]

Integration of 3D Geometric Distortion Assessment Into An Automated Daily End-To-End MRL QC Tool

D Mateescu, V Malkov*, A Damyanovich, S Jelveh, J Winter, D Letourneau, Princess Margaret Hospital, Toronto, ON


SU-K-BRC-5 (Sunday, 7/10/2022) 5:00 PM - 6:00 PM [Eastern Time (GMT-4)]

Ballroom C

Purpose: Geometric distortion can cause inaccuracies in patient external, target, and organ definitions in MR-guided radiation therapy. In this work we implement a geometric distortion assessment integrated into an automated daily end-to-end (E2E) quality control test on the Elekta Unity MR-linac.

Methods: The MR-to-MV phantom, containing seven MR/MV-visible fiducials, was used daily for E2E testing. The phantom was placed near the isocenter followed by a 2-minute T1-weighted image acquisition. To evaluate geometric distortion, we localized the phantom center, determined the positions of the fiducials, and obtained the distance between these and the phantom center. The distances were then compared to baseline values, established via MV imaging.Geometric distortion was evaluated for 398 MR images collected with our E2E test and compared with the geometric distortion measured weekly using the large Philips grid phantom. We validated MR and MV automated fiducial detection by comparison with manual localization and demonstrated accuracy in the presence of distortion by applying varying synthetic distortions to the MR phantom images (N=20) and repeating the distortion analysis.

Results: Geometric distortion measurements compared well with the grid phantom results (average differences < 0.5 mm). Variation from individual fiducial baselines was <0.25 mm and the magnitude of all distortions was <1 mm. MV detection by our algorithm differed from manual localization by at most -0.04 ± 0.08 mm, whereas MR detection differed by 0.20 ± 0.07 mm for the phantom center and -0.08 ± 0.20mm for individual fiducials. Synthetic distortions as large as 2.33 mm were detected to within 0.14 ± 0.02 mm.

Conclusion: Our work has demonstrated the accuracy of daily 3D geometric distortion measurements using the MR-to-MV phantom as part of our E2E QC workflow. These measurements are useful in that they allow automated daily assessment and quantification of image distortion over a clinically relevant 3D field-of-view.

Funding Support, Disclosures, and Conflict of Interest: This work is supported by a grant from Elekta


Geometric Distortion, MRI, Quality Assurance


IM/TH- MRI in Radiation Therapy: MRI/Linear accelerator combined Quality Assurance

Contact Email