Dynamic Computed Tomography (CT) emerges as a pivotal imaging modality for the assessment of knee joint kinematics. However, integrating dynamic CT into clinical practice necessitates a thorough understanding of healthy knee kinematics, as large variation in kinematics has been described within healthy populations. Therefore, this study aims to identify and describe healthy phenotypes with homogenous knee kinematics using a clustering approach. A total of 120 healthy knees from 64 participants underwent dynamic CT scanning during knee extension and flexion. Eight tibiofemoral (TF) and patellofemoral kinematic parameters were extracted, after which K-means clustering was applied to identify homogenous kinematic clusters. Kinematic phenotypes were obtained by calculating the median and interquartile range (IQR) for all kinematic parameters per cluster. Two distinct clusters were found, comprising 53 (Cluster 1) and 67 (Cluster 2) knees. Statistically significant differences between the clusters were found in six out of eight kinematic parameters. The most notable differences were observed in TF rotations, with cluster 1 exhibiting a greater amount of internal and adduction rotation of the tibia compared to cluster 2. The two kinematic phenotypes provide new insights into the nuanced variation within a healthy cohort and can serve as reference for future studies evaluating pathological kinematic phenotypes using dynamic CT.
Read full abstract