3D KINEMATIC PATTERN CLASSIFICATION OF HEALTHY KNEE JOINTS AND COMPARISON WITH OSTEOARTHRITIS JOINTS

Abstract

Purpose: To determine if some subsets of healthy subjects displayed other than a typical gait pattern and to identify which subsets have similar kinematic pattern to patients with knee osteoarthritis.

Methods: The healthy subject dataset consisted of 106 asymptomatic volunteers. These subjects were over 17 years of age, pain-free, had no record of surgery to the lower limb and no evidence or history of arthritic disease at the time of testing. The patient population consisted of 12 patients diagnosed with knee OA, evaluated within 6 months prior to the tests. The 3D movements of right knee joint were recorded using a functional knee analyzer with magnetic sensors while subjects walked on a treadmill at their own preferred speed. The magnetic sensors are non-invasive electromagnetic devices, which track the 3D positions and orientations of sensors relative to a source. The system has been shown to be accurate, especially in the frontal and transversal planes. K-means clustering analysis was chosen to identify the gait patterns among healthy subjects based on three components of the knee joint angles, and analyses of variance were performed to determine which parameters were different between subsets.

Results: Three gait groups or patterns were identified in the healthy subjects. The first group (G1) was characterized by a kinematic profile similar to the OA group. The second group (G2) had the highest external rotation angle, which was significantly different from OA group. The abduction angles were always greater in the G2 and G3 than in the OA group. This might be attributed to a valgus static alignment in G2 and G3 comparing to a varus alignment in the patient with OA.

Conclusions: The newly developed functional knee analyzer provided a non-invasive way to accurately measure 3D kinematic data which enabled cluster analysis to distinguish three gait patterns from 106 healthy subjects. The results suggested a strong correlation between static alignment and dynamic ad-abduction angles during the gait, which need to be investigated. Funding: Other Education Grant Funding Parties: NSERC, CIHR and FCAR