Three-dimensional computerized reconstruction. Illustration of incremental articular cartilage thinning.

RATIONALE AND OBJECTIVES: The authors have addressed the ability of magnetic resonance (MR) imaging to resolve incremental thinning of articular cartilage by assessment of three-dimensional (3-D) and two-dimensional (2-D) representations. METHODS: Using a porcine knee model, sequential cartilage shavings were characterized using a 3-D fat suppressed spoiled gradient-echo (SPGR) MR imaging protocol that provided good contrast between high-signal articular cartilage and lower signal surrounding tissues. Lesion dimensional measurements were made on both MR images and 3-D computerized reconstructions. Volumes of cartilage removed were approximately 0.06 mL. RESULTS: Incremental articular cartilage thinning typically was apparent on 3-D reconstructed images. Three-dimensional articular cartilage reconstructions were effective in depicting location and orientation of shaved cartilage regions. Average percent error associated with length and with measurements based on 2-D MR images was approximately 19% for observer 1 and 33% for observer 2 when compared with direct measurements of the shaved cartilage. Average percent error of thickness measurements based on 2-D MR was approximately 21% for observer 1 and 37% for observer 2. Overall average errors associated with length, width, and thickness measurements were approximately 25%. CONCLUSIONS: Incremental thinning of articular cartilage can be tracked qualitatively and quantitatively using 3-D computerized reconstructions and 2-D MR images. Errors associated with the quantitative measurements can be attributed to limitations of measurement methods and intrinsic limitation of MR resolution.

Analysis of three-dimensional computerized representations of articular cartilage lesions.

RATIONALE AND OBJECTIVES: The authors assess the accuracy of three-dimensional (3D) computer representations based on magnetic resonance images of articular cartilage lesions, using actual cartilage lesions as reference standards. METHODS: Grade 2 and grade 3 articular lesions were created on articular surfaces of five porcine knee joints. The knees were then imaged using 3D fat-suppressed SPGR acquisition at four different slice thicknesses. Magnetic resonance imaging data sets were transferred to a computer workstation for image processing and 3D reconstruction. Lesion dimensions (length, width, and depth) based on the 3D reconstructed image were compared with the dimensions measured using actual lesions. RESULTS: The average percent error of lesion length, width, and depth based on the 3D images ranged from approximately 8% to 12% when using the thinnest magnetic resonance slice thickness (0.7 mm). CONCLUSIONS: Three-dimensional reconstructed images derived from thin-slice magnetic resonance imaging can provide reasonable representations of true articular cartilage lesion dimensions.