Characterization of a Novel Imaging-Based Metric of Patellofemoral Separation Using Computational Modeling

Abstract

We introduce patellofemoral separation (PFS) as a novel metric to quantify patella-trochlear proximity as a function of dynamic knee flexion. PFS is quantified in 4D (i.e. 3D+time) using accurate segmentation from pre-operative imaging data acquired in three discrete, quasi-static knee postures, up to the maximum bending limit (i.e. 40° of flexion), within the constraints of a standard computed tomography (CT) or magnetic resonance imaging (MRI) scanner. Additionally, in this study, in order to examine patient-specific patella postures over a full range from 0 to 90° of dynamic knee flexion and extension, we utilize a computational model to simulate dynamic patella kinematics beyond 40° of bending. The computational model was optimized to reproduce patella postures as determined from the imaging data. A method of shape-based interpolation of the acquired 3D components (i.e. bone and cartilage) of the knee was applied in order to recreate a continuous range of motion of the patella and femur during knee bending from 0° to 40° using imaging data and 0° to 90° from simulated data. Next, a regional Hausdorff distance mapping paradigm was applied to compare the separation of the 3D surfaces defined by the patella and femoral cartilage segmentations from the interpolated imaging-based and simulated knee postures, at 1°increments. This separation distance was termed as PFS and examined as a posture-varying color map on the patella cartilage surface. The mean PFS was computed as the mean HD of separation between patella and femoral cartilage, at each posture over the entire studied range of motion. Mean PFS was observed to decrease with increased knee flexion, evidencing increased proximity of the patella and femur and increased risk of contact. In order to automatically quantify signs of patellofemoral instability from pathological knee kinematics reconstructed using medical imaging, the limits of PFS defining the thresholds of pain will require to be determined by benchmarking the metric against patients with normal knee-function. The PFS metric may also find potential application as a biomarker for the identification of high localized patellofemoral pressure by predicting patellofemoral impingement.