ObjectivesTo identify an optimal pediatric musculoskeletal ultrasound (MSUS) protocol for the detection of knee arthritis in patients with juvenile idiopathic arthritis (JIA) including a comparison with existing protocols. Secondary aims were to correlate MSUS-identified B-Mode (BM) and Power Doppler-Mode (PD) synovitis with clinical findings.MethodsConsecutive JIA patients with confirmed knee arthritis after clinical examination underwent a thorough MSUS study protocol which included views identified and consented by the Pediatric Rheumatology european Society (PReS) Imaging Working Party for the detection of synovitis. In total eight views including measurement of the suprapatellar recess were included. Scoring of synovitis followed the pediatric OMERACT criteria (BM and PD severity grading 0 to 3). Interobserver reliability of BM and PD was tested before study begin. Previously published MSUS protocols for knee synovitis were also identified from the literature and their scan protocols compared to identify differences in sensitivity for synovitis according to the number and specific type of views included. Finally, a clinically applicable MSUS protocol for knee synovitis could be proposed.ResultsIn 114 patients with clinically active knee inflammation, BM positivity (grading ≥ 1) was most frequently detected in the suprapatellar longitudinal and transverse scans performed in any positioning (frequency 97–99% in suprapatellar longitudinal in 30° or neutral respectively). PD positivity was however higher in these views performed in 30° flexion compared to neutral. Intrasynovial PD positivity (grading ≥ 1) was most frequently detected in the lateral parapatellar (69%, sensitivity 0.68, specificity 0.98), medial parapatellar (frequency 67%, sensitivity 0.67, specificity 1.0), the longitudinal lateral (68%, sensitivity 0.67, specificity 0.98) and suprapatellar transverse in 30° (frequency 64%, sensitivity 0.64, specificity 1.0). A combination of five views was the most sensitive for BM and PD synovitis. The suprapatellar recess size was analyzed by age and gender. For each group, the recess was wider in knees with arthritis than without (p < 0.001). Interobserver reliability of BM and PD positivity showed 85% agreement, with kappa 0.74 (very good). Three published studies with knee synovitis MSUS protocols were identified, which included a range of 1–3 views. Evaluation of the sensitivity of positive PD findings of each of these protocols reached a range of 53–83%; the highest sensitivity (91%) was achieved with the 5 views as identified by this study. These five views were therefore combined to form the Pediatric Internationally agreed Ultrasound (PIUS) knee protocol.ConclusionBM and PD positivity reliably correlated with the identification of pathological findings in knees of patients with JIA. From an internationally agreed protocol of eight images, a combination of five showed the greatest sensitivity for synovitis. This protocol, termed ‘PIUS-Knee’ performed well when compared to existing protocols.