Abstract
INTRODUCTION
Motion analysis is routinely used in the clinical and research sectors to quantify joint biomechanics. It plays an important role in clinical assessments by aiding the physician to distinguish between primary movement abnormalities and any secondary compensatory mechanisms that may overshadow the cause of the problem. During a data collection session, a wealth of biomechanical data regarding joint and segment kinematics and kinetics are collected from patients performing daily activities. Objective classification can be used to automate a diagnosis from this data and has been used previously to analyse measurements of level gait [1]. It is of interest to assess the knee during stair-gait as this activity involves greater range of motion (ROM) of the lower limbs, larger forces and moments acting at the knee.
AIM
The aim of the current study is to explore the use of an objective classifier [1] to characterise knee osteoarthritis (OA) and monitor functional recovery following a total knee replacement (TKR) using measurements from stair-gait.
METHODS
Motion analysis techniques were used to quantify knee OA kinematics and kinetics during stair-gait for six patients with knee osteoarthritis (OA) and nine subjects without pathology (NP). One OA subject, forming a TKR sample, was also assessed at 4, 8 and 12 post-operatively. Each subject performed three trials of stair ascent and descent. 3D motion capture was performed using 8 Qualisys MCUs, capturing at 60Hz and a 1000Hz force plate (Bertec Corporation). Forces were measured from the first step of the staircase [2]. Independent t-tests were performed on biomechanical measures to compare the NP and OA cohorts (p<0.05). This identified the adaptations associated with knee OA. Principal components of salient kinematic and kinetic waveforms were used as inputs to train the classifier and subsequently characterise recovery of the TKR sample.
RESULTS
The OA cohort adapted their stair-gait by reducing their peak: (i) external flexion moment in stance during both stair ascent and descent; (ii) medial ground reaction force (GRF) (iii) vertical GRF during stair descent and increasing their external adduction moment during stair ascent. The classifier was used to characterise knee function of the OA and NP subjects with 100% classification accuracy, defined using a Leave-one-out cross-validation. The TKR sample was classified as having dominant OA functional characteristics pre-operatively. At all subsequent measurements the subject was classified as having NP stair-gait characteristics. These changes correlated significantly with Knee Outcome Survey and Oxford Knee scores.
CONCLUSION
Classification is a powerful tool for characterising data into two groups where a simplex plot provides a simple clinical interpretation of the results from a motion analysis assessment. This study demonstrates the use of objective classification to quantify NP, OA and TKR function from stair-gait. It also demonstrates its capability to monitor functional changes during a subject's recovery.
