Cover Image

3D quantification of metal-induced geometric distortions in MRI

The increasing number of patients with metal implants raises concerns about metal-induced geometric distortions (MD) in MR-guided treatments. This study proposes a method for three-dimensional quantification of MD and evaluates its accuracy and reliability. A 3D lattice phantom was designed and meas...

Full description

Saved in:
Bibliographic Details
Main Authors: Li, Hao (Author) , Özen, Ali C. (Author) , Jürchott, Alexander (Author) , Breckwoldt, Michael O. (Author) , Jesser, Jessica (Author) , Vollherbst, Dominik (Author) , Schwarz, Daniel (Author) , Bendszus, Martin (Author) , Heiland, Sabine (Author) , Hilgenfeld, Tim (Author)
Format: Article (Journal)
Language:English
Published: 28 February 2025
In: Scientific reports
Year: 2025, Volume: 15, Pages: 1-12
ISSN:2045-2322
DOI:10.1038/s41598-025-90645-5
Online Access:Verlag, kostenfrei, Volltext: https://doi.org/10.1038/s41598-025-90645-5
Get full text
Author Notes:Hao Li, Ali C. Özen, Alexander Juerchott, Michael Breckwoldt, Jessica Jesser, Dominik F. Vollherbst, Daniel Schwarz, Martin Bendszus, Sabine Heiland & Tim Hilgenfeld
Description
Summary:The increasing number of patients with metal implants raises concerns about metal-induced geometric distortions (MD) in MR-guided treatments. This study proposes a method for three-dimensional quantification of MD and evaluates its accuracy and reliability. A 3D lattice phantom was designed and measured with two sequences (VIBE and SPACE) and two implants (crown-supported-dental-implant and stainless-steel-bracket). Automated detection of displacement of 9360 crossing points caused by MD was performed. Distortion-quantification accuracy was improved by correcting for noise-induced error (NE), related to different signal-to-noise ratios (SNR), and implant-related signal loss and pile-up artifact volumes (SLPUA). The method’s accuracy was validated against computed tomography. Results showed high reliability, with an excellent intraclass correlation coefficient (≥ 0.99) and low mean residual errors in all directions (2.6%/1.6%/1.8% of voxel size in X/Y/Z direction). SNR/SLPUA volumes were significant confounders (p-value ≤ 0.001) when comparing different sequences/implants, but corrections significantly reduced their impacts (p-value ≤ 0.001). This method enables accurate 3D MD quantification and fair comparison across different sequences/implants. By optimizing MRI protocols for MD minimization and defining implant-specific MD profiles for patient data correction, it may help improve spatial accuracy in MRI-guided treatments in the future.
Item Description:Gesehen am 09.10.2025
Physical Description:Online Resource
ISSN:2045-2322
DOI:10.1038/s41598-025-90645-5

Similar Items