Author: Emel Calugaru, Jenghwa Chang, Sean T Grace, Nicholas Harvey, Jessica Jung, Ching-Ling Teng 👨🔬
Affiliation: Northwell, Hofstra University, Hofstra University Medical Physics Program 🌍
Purpose: To evaluate the consistency of MRI distortion corrections across Northwell Health system and evaluate the necessity and importance of distortion QA for MRIs used for stereotactic radiosurgery (SRS) regardless of manufacturers’ scanning algorithms.
Methods: The Quasar MRI Grid phantom was scanned with the distortion correction algorithm applied at seven different MRI units with a total of 17 scans were obtained following a standardized localization protocol. The reference image was set by three different methods: (1) localization box alignment and (2) utilization of the “ideal matching” provided by the Modus QA analysis software. CT localization box with the MRI localization box alignment accuracy was quantified using mean and maximum stereotactic reference values calculated in GammaPlan after stereotactic body reference. No standardized method was available to assess the accuracy of the CT fusion, introducing additional uncertainty and the ideal algorithm provided direct comparison metrics with the ideal reference.
Results: Across all sites, the ideal algorithm consistently demonstrated the lowest deviation from the reference positions, with average absolute deviations of 0.306 mm in the x, y, and z coordinates. Localization box alignment yielded a mean stereotactic error of 0.2 mm and a maximum error of 0.3 mm. Fusion to CT scans showed the highest variability, underscoring the need for a standardized validation protocol. Variations in distortion between sites were observed, with site-specific deviations ranging from 0.15 mm to 1.5 mm.
Conclusion: Significant distortion (up to 1.5 mm) was still observed for stereotactic MRI scans across our health system even correction algorithms were applied, highlighting the need for additional distortion QA for our SRS program. The “ideal algorithm” is the most reliable method for standardizing the QA process for MRI distortion using the Grid phantom across multiple sites.