Search

Objective: MRI of small joints plays an important role in the early detection and early treatment of rheumatoid arthritis. Despite its sensitivity to demonstrate inflammation, clinical use is hampered by accessibility, long scan time, intravenous contrast, and consequent high costs. To improve the feasibility of MRI implementation in clinical practice, we introduce a modified Dixon sequence, which does not require contrast and reduces total acquisition time to 6 min. Because the reliability in relation to conventional MRI sequences is unknown, we determined this. Methods: In 29 consecutive early arthritis patients, coronal and axial T2-weighted modified Dixon acquisitions on 3.0 T MRI scanner were acquired from metacarpophalangeal 2–5 to the wrist, followed by the standard contrast-enhanced protocol on 1.5 T extremity MRI. Two readers scored osteitis, synovitis and tenosynovitis (summed as total MRI-inflammation), and erosions (all summed as total Rheumatoid Arthritis MRI Score (RAMRIS)). Intraclass correlation coefficients (ICCs) between readers, and comparing the two sequences, were studied. Spearman correlations were determined. Results: Performance between readers was good/excellent. Comparing modified Dixon and conventional sequences revealed good/excellent reliability: ICC for total MRI-inflammation score was 0.84 (95% CI:0.70–0.92), for erosions 0.90 (95% CI:0.79–0.96), and for the total RAMRIS score 0.88 (95% CI:0.77–0.94). The scores of total MRI-inflammation, total erosions, and total RAMRIS were highly correlated (ρ = 0.80, ρ = 0.81, ρ = 0.82, respectively). Conclusion: The modified Dixon protocol is reliable compared to the conventional MRI protocol, suggesting it is accurate to detect MRI inflammation. The good correlation may be the first step towards a patient-friendly, short and affordable MRI protocol, which can facilitate the implementation of MRI for early detection of inflammation in rheumatology practice. [ABSTRACT FROM AUTHOR]