2089 - Toward Developing an MRI QA Program for SRS and fSRS Treatments within our Multicampus Radiation Medicine Center

Purpose/Objective(s): We have a well-established non-invasive stereotactic radiosurgery program since 2014 that serves all patients of the health system who need to receive brain SRS and fSRS treatments. Since most of our GK cases are planned without an added margin, image quality and geometric accuracy of the planning MRI scans are critical to the success of our GK treatments. Due to the geological spread of our clinics, the MRI scans for our GK program are acquired from various units across our health system. Sharing MRI Units of various diagnostic radiology departments for GK program presents unique challenges as these MRI units are QAed by different Diagnostic Physics Teams without a centralized quality check. Therefore, an internal dedicated GK MRI QA Protocol is urgently needed to ensure high accuracy and precision in MRI imaging techniques for our GK patients.

Materials/Methods: Based on the recommendations in AAPM TG284 and TG178, we have performed an initial evaluation of image quality and geometrical accuracy of the MRI units across our health system used for GK planning. The 3D GRID MRI PHANTOM and the ACR MRI PHANTOM were scanned at 7 MRI Centers, with the GK scanning protocol (3D distortion correction algorithm on, 256x256, slice 1mm) and GK sequences: T1MPRAGE, T1SPGR, T2, T2 FRFSE, T2 CISS. A total of 27 scans of the 3D GRID PHANTOM were obtained for each MRI unit and each scan was analyzed for 3D distortion as function of site, sequence, coil, localization box. A total of 8 scans of the ACR PHANTOM were obtained and assessed for image quality check including high contrast spatial resolution, image intensity uniformity, percent signal ghosting, and low contrast object detectability.

Results: For image quality QA, all analysis passed but the high contrast spatial resolution varied between .9 and 1mm. For MRI distortion QA, distortion differences were observed between sites, with site-specific deviations ranging from 0.15 mm to 2 mm in x-, y- and z-direction as well as along a 3D radius of 5cm from the DICOM origin. Additionally, significant distortion, reaching up to 2mm, persists in stereotactic MRI scans across our health system despite the application of correction algorithms. This highlights the need for enhanced distortion QA within our SRS program. Among the correction methods evaluated, the "ideal algorithm" proved to be the most reliable approach for standardizing MRI distortion QA using the Grid phantom across multiple sites.

Conclusion: Our preliminary QA check demonstrated a high variation of MRI distortion among the MRI units across our health system, even though every imaging center applied the vendor-recommended distortion correction algorithm. This study underscores the necessity for dedicated SRS/fSRS MRI QA protocols to ensure accuracy and consistency across sites. We plan to investigate the passing criteria for these QA checks to further achieve consistent distortion and uniform image quality.