ESTRO 2024 - Abstract Book

S4767

Physics - Quality assurance and auditing

ESTRO 2024

662

Digital Poster

Implementation of an MRI QC for SRS purposes

Noémie Johnston 1 , Eric Lenaerts 1 , Gautier Kempinaire 2 , Véronique Baart 1

1 CHU de Liège, Radiotherapy, Liège, Belgium. 2 CHU de Liège, Radiodiagnostic, Liège, Belgium

Purpose/Objective:

All brain SRS patients undergo a simulation CT for dose calculation and structure delineation, as well as an MRI to assist with the structure delineation. However, MRI is subject to distortions, caused among others by static field inhomogeneity and gradient field non-linearity. Such distortions can lead to erroneous delineation of the target or organs at risk. During 2022 a new 3T GE Architect MRI was installed, on which radiotherapy patients are now being imaged. Patients with pacemakers that are not 3T compatible are imaged on a 1.5T Siemens Sola MRI. As a result, we reviewed our quality assurance of the MRI and implemented a distortion quality control.

Material/Methods:

We followed the NCS36 1 and the TG 284 2 recommendations for the elaboration of the tests and its tolerances. The tests can be separated in an image quality and distortion control. The image quality control includes SNR, uniformity, resonance frequency, spatial resolution, low contrast detectability, slice position, slice thickness and ghosting. We followed the American college of radiology (ACR) MRI accreditation protocol, using their large ACR phantom. On top of the ACR suggested sequences, we checked the image quality on the radiotherapy brain sequence 3D T1 SPGR Tra. Images were acquired monthly starting April 2023 and were analysed using ImageJ. For the distortion QC, we used the QUASAR MIRD3D phantom and software. The sequences used were a 3D T1 FSPGR 1.5mm acquisition and a 3D T1 SPGR Tra acquisition, we compared the sequences with and without the distortion correction. We compared the distortions on the Ge Architect MRI to the one on the Siemens Sola MRI. Images were acquired bi-monthly since June 2023.

Results:

The image quality results were quite variable, especially the SNR results. For our clinical SRS sequence, all tests passed successfully. This was not the case for the ACR sequences. Namely, the percent image uniformity was inferior to 50. In a 32cm diameter sphere, centred at isocentre, the mean distortion was 0.262 ±0.039mm in all three directions (x,y,z), with the max distortion being inferior to 2mm. If the built-in distortion correction was not applied, the mean distortion along the x and y direction did not significantly change. However, the distortions along the z-axis increased significantly, with in average, a mean distortion of 1.773 ±0.007mm and a maximum of 8.943 ± 0.090mm. No significant differences were observed between the 3T GE Architect MRI and the 1.5T Siemens Sola MRI.

Conclusion: