Session Item

It is known that MRI is not as geometric accurate as CT. In MRI-guided RT this can have implications for the accuracy of dose delivery. MRI is now available in all RT facilities in Denmark and to monitor image quality, a national network for quality assurance of MR images in RT in Denmark (NIMBUS) was founded in 2018. The aim of the network is to improve quality by performing and sharing measurements and knowledge across all centers in Denmark and in particular to secure image quality in national clinical trials. By identifying baseline performance for MR image quality and quantitative MRI on a national level, centers that deviate can be identified and action can be taken to improve quality. In this study, the geometric accuracy of different MR scanners in eight centers was evaluated.

The steering committee of NIMBUS reached out to all RT centers in Denmark through an MRI QA workshop to launch the network. At the participating centers, a general MRI sequence (3D T1W GRE) was used to test geometric accuracy with adjustments in repetition time and echo time to accommodate differences in field strength. A large field of view phantom (Magphan RT 820) was used in a traveling-phantom setup for all acquisitions. The acquisitions were made in the period from March 2019 to December 2020.

Distortion analysis of the MRI scans was performed in the web-based software Smári Image Analysis Service (The Phantom Laboratory, NY, USA). The analysis output was distortions as a function of distance to iso-center as well as maximum distortion and mean of the 10% maximum distortions within a 200 and 350 mm diameter spherical volume (DSV). The mean of the 10% maximum distortions was reported as a robustness check of the maximum distortion.

Seven of eight Danish RT centers and one radiological department have participated in this study so far. In total, the centers had eight MRI scanners and two MR-linacs (Table 1) which were evaluated.

All maximum distortions were below 0.43 mm within 200 mm DSV and below 1.2 mm within 350 mm DSV, except for the Siemens Aera 1.5 T (Figure 1). The mean of the 10% maximum distortions was in general about 20% lower than the maximum reported values; for all scanners, except for the Siemens Aera 1.5 T, the mean of the 10% maximum distortions was less than 1 mm within the 350 mm DSV. No dependency on field strength was seen for the reported distortions. 

In general, the scanners at the participating centers showed maximum distortions below 0.5 mm within 200 mm from the iso-center of the scanner. The reason for the high level of distortion of the MRI scanner at one center has not yet been found, but the detection of an outlier proves the value of the national collaborative MRI quality assurance network. The next steps include analysis of clinical MRI sequences, other image quality metrics, independent distortion calculation, and a web-based quality assurance system for monitoring.