Methods. Forty-one patients and 12 healthy controls participated in this cross-sectional study. Coronal short tau inversion recovery (STIR) and T1-weighted sequences were obtained at 0.23, 0.6, 1.5 and 3T using flex coils (Flex). Additional STIR sequences were obtained with phased array extremity coils (Extr) (at 0.6 and 1.5T) and higher resolution (at 1.5T). In otal, 338 STIR image sets were anonymized and scored according to RAMRIS and parameters of image quality were measured.
Results. The BME sum scores were similar overall when comparing the different MRI units, coil types and voxel sizes, yet significantly higher at the higher resolution of 1.5T Extr compared with 0.23T Flex (P = 0.004), 0.6T Flex (P = 0.03), 1.5T Flex (P = 0.05) and 3T Flex (P = 0.001). Mean differences were relatively minor (0–3.5). Intrareader reliability of BME scores was high [intraclass correlation coefficient ≥ 0.90 for all except 0.23T (0.81) and percentage exact agreement 81–88%]. The smallest detectable difference was better at 0.6, 1.5 and 3T (9–29% of maximum value) than at 0.23T (40%). Image quality was lowest at 0.23T.
Conclusion. No major, consistent differences were found between BME scores using STIR sequences obtained at different field strengths, coil types and image resolutions, suggesting that these are equally suited for assessment of BME in RA. However, parameters of image quality and intrareader reliability (favouring 0.6, 1.5 and 3T) should be considered when selecting the MRI acquisition strategy.