OBJECTIVES: To explore if the reliability of synovitis assessment by unenhanced MRI is influenced by different MRI field-strengths, coil types and image resolutions in RA patients.

METHODS: Forty-one RA patients and 12 healthy controls underwent hand MRI (wrist and 2(nd)--5(th) metacarpophalangeal joints) at 4 different field-strengths (0.23 T/0.6 T/1.5 T/3.0 T) on the same day. Seven protocols using a STIR sequence with different field-strengths, coils (flex coils/dedicated phased-array extremity coils) and resolution were applied and scored blindly for synovitis (OMERACT-RAMRIS method). A 1.5 T post-contrast T1-weighted sequence was used as gold standard reference.

RESULTS: Fair-good agreement (ICC=0.38--0.72) between the standard reference and the different STIR protocols (best agreement with extremity coil and small voxel size at 1.5 T). The accuracy for presence/absence of synovitis was very high per person (0.80--1.0), and moderate-high per joint (0.63--0.85), whereas exact agreements on scores were moderate (0.50--0.66). The intrareader agreement (15 patients and 3 controls) on presence/absence of synovitis was very high (0.87--1.0).

CONCLUSIONS: Unenhanced MRI using STIR sequence is only moderately reliable for assessing hand synovitis in RA, when contrast-enhanced MRI is considered the gold standard reference. Contrast injection, field strength and coil type influence synovitis assessment, and should be considered before performing MRI in clinical trials and practice.

KEY POINTS: • STIR is only moderately reliable for synovitis assessment, compared with post-contrast-T1-w. • Contrast injection, field strength, and coil type influence synovitis assessment. • Contrast injection is recommended for reliable and reproducible hand synovitis assessment.