Abstract
OBJECTIVE: To validate a semi-automated software method of quantifying knee osteoarthritis (KOA) related effusion-synovitis (ES) and Hoffa-synovitis (HS) on MRI. MATERIALS AND METHODS: 301 subjects were randomly selected from the FNIH sub cohort, a nested case control study within the Osteoarthritis Initiative (OAI), and distributed into 4 groups based on pain and radiographic progression. Measurements of ES and HS volume were made by 2 readers. Criterion validation was assessed through comparison with the MRI Osteoarthritis Knee Score (MOAKS) and the Spearman correlation coefficient r value. Reader reliability was measured on a subset of 30 subjects and intra-class correlation coefficients (ICCs). Clinical validity was assessed based on case control status using logistic regression and the area under the curve (AUC). RESULTS: ES volume was highly correlated with MOAKS Scores (r = 0.74), as was the HS measure but to a lesser extent (r = 0.55). For ES, the intra-reader and intra-reader precision ICCs were 0.83 and 0.95 respectively and 0.98 and 0.96 for HS. For clinical validity, we found similar AUC values when comparing the software method to MOAKS. The average reader time was less than 15 min per knee for both ES and HS. CONCLUSION: We have demonstrated the validity of an efficient, accurate, and rapid ES and HS measurement method for KOA using MRI. To our knowledge, this is the first such software to measure both ES and HS. This method will offer an objective and efficient tool for clinical trials and other epidemiologic studies of KOA.