Key points are not available for this paper at this time.
PURPOSE: To determine the accuracy of semiautomated quantification of articular cartilage volume from three-dimensional (3D) reformations of magnetic resonance (MR) images. MATERIALS AND METHODS: Sagittal, fat-suppressed, 3D, spoiled gradient-recalled-echo MR imaging of two bovine and two human cadaver knees was performed. Articular cartilage volume was calculated from 3D reformations of the MR images by using a semiautomated program written at the authors' institution. Calculated volumes were compared with directly measured volumes of the surgically removed articular cartilage. RESULTS: The percentage of error of the MR imaging-determined volumes was 6.53% +/- 4.75 (mean +/- standard deviation). A strong correlation between the two sets of observations was shown (r=.997). Linear regression showed the calculated volumes to be highly accurate (slope=1.002, P>.25). Repeated reformations yielded volumes that were reproducible (mean absolute error, 0.013 mL +/- 0.019) and not significantly different from the measured volume (P>.10). CONCLUSION: Semiautomated quantification of knee articular cartilage from MR images yields highly accurate cartilage volumes.
Piplani et al. (Fri,) studied this question.