Purpose: Magnetic resonance imaging (MRI) is being used increasingly as a primary imaging tool for evaluating sacroiliac joints (SIJ) in juvenile idiopathic arthritis (JIA). Due to the need for the standardization of SIJ MRI interpretation, JIA MRI scoring system (JAMRIS-SIJ) has been developed. We aimed to evaluate the reliability of JAMRIS-SIJ since related studies are scarce in the literature.

Methods: Five readers─-one pediatric radiologist, two musculoskeletal radiologists, one general radiologist, and one radiology resident---blinded to the patients’ data, rated the MR images individually according to the JAMRIS-SIJ. The study comprised 40 MR examinations of 31 patients ≤18 years old with enthesitis-related juvenile spondyloarthritis. All available examinations contained T1-weighted (T1W), fat-suppressed T2-weighted (T2W), or Short Tau Inversion Recovery (STIR) sequences. Of them, 34 examinations also had fat-suppressed T1W, and 35 had contrast-enhanced imaging. The interrater reliability was detected with the interclass correlation coefficient (ICC).

Results: The interrater reliability ICC of JAMRIS-SIJ for the active inflammation was 0.870. Two different ICCs were calculated for the structural damage, one for T1W sequence and one for fs T1W sequence; the results were 0.654 and 0.724, respectively. Among the 14 components scored, only bone marrow edema and fat metaplasia had good reproducibility.

Conclusion: The reliability of JAMRIS-SIJ was good for the active inflammation. However, most of the items of both active inflammation and structural damage domains had poor reproducibility. Therefore, each item should be reconsidered according to their weightings.