LAUSR

OBJECTIVE Temporomandibular joints (TMJs) frequently develop silent inflammatory and osteochondral changes in children with juvenile idiopathic arthritis. Data-driven recommendations for TMJ imaging protocol are needed to reduce measurement error and scanning time. This study compares the impact of different protocols, imaging coils, and magnet strength on the reliability of image assessment and the subjective quality of images. MATERIALS AND METHODS Three groups of bilateral TMJ MR studies were retrospectively collected from two institutions, including 24 1.5T and 19 3.0T studies using dedicated TMJ surface coils, and 23 1.5T studies with head coil. Post-contrast sequences were re-compiled from the full protocol to create minimum protocol studies for the three groups. Two radiologists and two non-radiologists first scored the three minimum protocol images according to pre-specified definitions, then scored the full protocol images. Minimum-to-full protocol agreement, inter-reader agreement, and subjective item visibility scores were assessed. RESULTS With dedicated TMJ dual surface coils, minimum-vs-full protocol agreement was moderate to good (0.5-0.8 intraclass correlation coefficients or kappa) for most items, and was not influenced by the magnet strength. Inter-reader reliability was more significantly influenced by the imaging coil and reader's training background than by protocol length or magnet strength differences. Sagittal and coronal planes weighted on PD, T2 Fat Suppressed and T1 Fat Suppressed -postcontrast adequately visualized all the different features, whereas the axial plane was more limited to visualizing synovium. CONCLUSION Inter-reader reliability and qualitative measure of image quality improved more consistently with the coil offering the higher resolution, rather than increased magnet strength.