Purpose: To assess the reproducibility and accuracy of volumetric quantifications of articular cartilage in the knee determined with three-dimensional (3D) magnetic resonance (MR) imaging combined with pulsed saturation transfer subtraction (STS) or T1-weighted fat suppression (FS) imaging.
Materials and methods: Eight osteoarthritic knees were imaged repeatedly with optimized STS and FS sequences. Cartilage volumes were determined from 3D reconstructions of FS and STS images and by means of water displacement of surgically retrieved tissue.
Results: Mean over- or underestimation of cartilage volume at STS and FS imaging was 0.40 mL +/- 0.11 (standard deviation) (8.2%) and 0.31 mL +/- 0.08 (5.9%), respectively. Intraobserver reproducibility error was 0.20-0.65 mL (3.6%-6.4%) for STS and 0.21-0.58 mL (4.2%-6.4%) for FS imaging. Interobserver error was less than 0.62 mL and 7.8%.
Conclusion: Three-dimensional data analysis of MR images acquired with STS or FS allows accurate and reproducible volumetric quantification of articular cartilage in the knee.