P283 ONE YEAR OAI PILOT DATA: LONG TERM VARIABILITY OF CARTILAGE MORPHOLOGY AS MEASURED BY MRI DESS

Abstract

twice was evaluated using pairwise differences and linear correlation between pairs of measures. The surface area estimate from automatic segmentation gave a linear correlation coefficient of 0.96 between pairs of estimates on the same knee, and the mean absolute pairwise difference was 2.9%. Conclusions: The cartilage surface area estimate obtained from the automatic cartilage segmentation method can separate healthy populations from OA populations with statistical significance (p = 0.014). Since the method is fully automatic it has full intra-scan reproducibility, and the results show that is also has high inter-scan reproducibility (absolute pairwise difference of 2.9%, linear correlation coefficient of 0.96). Because the method is fully automatic, reproducible and can separate healthy populations from OA populations using the cartilage area estimate obtained from the method, it may be a useful tool in clinical studies using MRI.