Knee Injuries

A total of twenty subjects (n=10 with BMI of 18.5–24.9 kg/m² and n=10 with BMI of 25–31 kg/m²) participated in this IRB approved study. No subjects had a history of knee injury or knee surgery. Subjects in each group were age- and sex- matched. Each subject was imaged twice in one day (at 8:00 AM and 4:00 PM) using a 3T MRI scanner (Trio Tim, Siemens Medical Solutions USA, Malvern, Pennsylvania) and an eight channel knee coil. Subjects were instructed to refrain from exercise or any strenuous activity prior to the morning scan. Subjects were scanned while lying supine, with the knee in a relaxed, slightly flexed position (26 (link)). A double-echo steady-state sequence (DESS, field of view: 15×15 cm, matrix: 512×512 pixels, slice thickness: 1 mm, flip angle: 25°, repetition time: 17 ms, echo time: 6 ms) was used to generate sagittal plane images of the knee (Figure 1) (23 (link), 26 (link)). Total scan time was approximately 9 min. Subjects were asked to perform their normal activities throughout the day. Subjects returned to the same facility at 4:00PM and were imaged immediately upon arrival using the same protocol. In addition, each participant wore a pedometer to provide a measure of the number of steps taken between MR imaging sessions.
The MR images were used to generate 3D models of each subject’s knee, including models of the femur, tibia, patella, and the corresponding articular surfaces (Figure 1) (23 (link)). In each image, the outer margins of the cortices and articular surfaces of cartilage were segmented using a solid modeling software (Rhinoceros, McNeel and Associates, Seattle, WA) (23 (link), 27 (link)). These curves were then used to generate 3D surface mesh models of each surface (Studio, Geomagic Inc., Research Triangle Park, NC) (Figure 1). AM and PM models of the femur, tibia, and patella were aligned using an iterative closest point technique so that thickness measurements and strain calculations could be made at the same locations in each model (23 (link)). A grid sampling system was used to characterize cartilage thickness in different regions of the joint (23 (link)). Specifically, cartilage thickness on the tibia was sampled over a grid of nine evenly spaced points spanning both the medial and lateral tibial plateaus, for a total of 18 points (Figure 1). Eleven points spanned the surface of the patella and six points were sampled on the subpatellar region of the femur. A total of 36 points were measured across the surfaces of the medial and lateral femoral condyles. Thickness was defined as the minimum distance from the articular surface to bone at each mesh point on the model (Figure 2). Thickness was averaged across each mesh point within a 2.5mm radius of the sampling point. In this fashion, diurnal strain was calculated using the ratio of the change in thickness (final thickness minus initial thickness) to the initial thickness at the same location (23 (link)). This methodology has been previously validated for measuring cartilage thickness in the tibiofemoral joint (28 (link)). A more recent study from our laboratory reported a coefficient of repeatability of 0.03mm; thus, we estimate the resolution of strain measurements to be less than 1% (23 (link)).