Healthy cartilage and subchondral bone are critical for normal joint function. The ability to non-invasively quantify the spatial geometry of articular cartilage (AC) and subchondral bone (SB) defects remains a challenge. Non-invasive magnetic resonance (MR) imaging may be valuable for longitudinal study of chronic osteoarthritis (OA), acute osteochondral injury, or chondroprotective therapies. However, longitudinal MR assessment of AC and SB thickness is limited due to accuracy and repeatability issues associated with out-of-plane curvature of joint surfaces. Computer models for three-dimensional (3-D) estimation of AC and SB thickness have been proposed, but not rigorously validated. The objective of this research is enable accurate, non- invasive, 3-D measurement and visualization of cartilage and bone focal defects resulting from disease or injury. Our basic hypothesis is: accuracy of assessing cartilage and bone thickness defects will be greatly enhanced with visually informative and quantitative 3-D thickness distribution data from co-planar MR images.
Our specific aims are: 1. To validate an analytical model for computer generation of AC and SB spatial thickness from MR images. Normal cadaver joints will be mounted on an MR-safe miter table during MR imaging, then sectioned using the miter table guides. AC and SB spatial thickness maps will be generated from the MR data and validated with optical measurements of anatomical sections. Hypothesis: out-of-plane errors in AC and SB thickness from MR imaging can be modeled and corrected to within voxel dimension at a 99 percent confidence level using the analytical model. 2. To test the model's performance for detecting and sizing AC and SB focal defects in cadaver joints. Defects of known diameter and depth will be created in normal joints and cartilage lesions in abnormal specimens will be assessed and measured. 3-D thickness maps generated from the MR data will be then used to locate, size and assess the defects. Hypothesis: the analytical model can exhibit defect detection sensitivity and specificity greater than 98 percent, with a dimensional accuracy of 1 voxel dimension in defect depth and diameter at a confidence level of 99 percent. 3. To test the model's performance for detecting and sizing AC and SB focal defects in living humans. Patients presenting for total knee replacement will be MR imaged prior to surgery using routine clinical protocols. Surgically removed tissues will be assessed according to Aim 1 and Aim 2. Hypothesis: the analytical model can predict AC and SB thickness within 1 voxel dimension at a 95 percent confidence level and exhibit defect detection sensitivity and specificity greater than 98 percent.
