Abstract
PURPOSE: To determine the diagnostic performance and inter-rater agreement for magnetic resonance imaging (MRI) signs of long head biceps tendon (LHBT) instability and overall, using conventional MRI for the diagnosis of biceps pulley lesions. METHODS: In this retrospective analysis, conventional MRIs were reviewed by 5 assessors for the presence or absence of biceps pulley lesions and 6 specific MRI signs. Diagnostic performance of pulley lesion and sign detection using MRI was tested using arthroscopy as the reference standard. Interobserver agreement was measured with Kappa statistics and diagnostic performance with sensitivity, specificity, negative and positive predictive values overall and for radiologists and surgeons. RESULTS: A total of 60 MRIs, 30 with biceps pully lesions and 30 without, were included. Overall, diagnostic performance metrics for MRI included a sensitivity of 81%, specificity of 79%, positive predictive value of 80%, and negative predictive value of 80%. Interobserver reliability analysis revealed moderate agreement overall, with a global kappa value of 0.59. LHBT angle showed the highest sensitivity (84%) and LHBT-groove distance showed the greatest specificity (98%). Both radiologists and surgeons reported similar diagnostic accuracy through MRI. CONCLUSIONS: Overall, conventional MRI had an acceptable diagnostic performance, with sensitivity, specificity, and predictive values of approximately 80%. Among the evaluated signs, the LHBT angle had the highest sensitivity, whereas the LHBT-groove distance showed the greatest specificity. Interobserver reliability was moderate overall, though some observer pairs achieved substantial agreement. However, there was variability across diagnostic signs. LEVEL OF EVIDENCE: Level III, retrospective comparative study.