Spatial characterization of signal-to-noise ratio (SNR) differences among multiple sites using a phantom

Motivation: The motivation came to acknowledge the aid for sequence standardization and to assess other site-related differences. Goal(s): We aim to consider the SNR differences to improve comparability between sites in structural analysis of brain data. Approach: We describe preliminary MRI phantom data from a longitudinal, multi-site MS study, illustrating an approach to sequence standardization and assessing site-related differences. Results: Scan-rescan data were collected on three 3T MRI systems, using study specific, previously optimized sequences. Signal-to-noise ratio (SNR) was measured on a 16 x 16 grid structure within the phantom. Imaging voxels within the grid where algorithmically identified and SNR assessed in each cube. Impact: Large between-site differences in SNR and spatial variability were observed, while longitudinal data showed good consistency in mean SNR and spatial appearance. These differences underscore the need for correction to improve between-site comparability.