Comparison between 3-point Dixon- and CHESS-based OMERACT-recommended MRI protocols in hands of patients with suspicion of early rheumatoid arthritis.

Kirchgesner, Thomas [UCL] Stoenoiu, Maria [UCL] Michoux, Nicolas [UCL] Durez, Patrick [UCL] Vande Berg, Bruno [UCL]

(eng) PURPOSE: To compare fat suppression effectiveness, image quality and disease activity scores between MRI protocols based on the Dixon method and the Chemical Shift Selective (CHESS) technique in hands of patients with suspicion of early rheumatoid arthritis (RA). METHOD: Both hands of 28 patients (19 women; mean age 45.2 years old) with suspicion of early RA were prospectively imaged with Dixon- and CHESS-based OMERACT recommended protocols at 1.5 T including fat-suppressed T2-weighted and contrast-enhanced T1-weighted imaging. Two radiologists (R1/R2) separately assessed effectiveness of fat suppression and determined RAMRIS scores woth the Dixon- and CHESS-based protocols. R1 repeated the RAMRIS scoring and measured contrast-to-noise ratios (CNRs) on Dixon and CHESS images. Statistics included 2-way ANOVA test for the comparison of CNRs and Bland-Altman methodology for inter-technique and intra-observer agreement (p < 0.05). RESULTS: Fat suppression failure occurred in up to 1 patient with the Dixon- and 25 patients with the CHESS-based protocols. CNRs were significantly higher on T1-weighted and lower on T2-weighted Dixon images than on the corresponding CHESS images (p ≤ 0.042). Median bias of the difference between Dixon- and CHESS-based RAMRIS scores was not significantly different from 0 (-0.8 to +1.0 and -1.1 to +1.4 for R1/R2). Median bias of the difference between RAMRIS scores at first and second readings was significantly different from 0 with the CHESS-based protocols (-0.8 to +1.7) but not with the Dixon-based protocols (+0.0 to +1.0). CONCLUSIONS: Dixon sequences yield more effective fat suppression and more reproducible RAMRIS scoring than CHESS sequences in hands with suspicion of early RA.