Purpose: Laboratory motion capture testing has established that older adults and individuals with knee osteoarthritis (OA) walk slower and take shorter and more frequent steps than their young counterparts. Additionally, individuals with knee OA have altered gait kinematics demonstrated by a smaller range of knee motion during walking. Early detection of subtle changes in gait could be critical in slowing deterioration of mobility and the progression of joint damage related to age and knee OA. Moreover, observation of walking dynamics outside a controlled setting may provide insight into discrepancies between walking capacity (measured in a gait lab) and walking behavior (free-living environment). Recent advances in wearable sensors, including inertial measurement units (IMUs), provide the opportunity to assess movement and gait outcomes outside of the laboratory setting. Before this real-world assessment can occur, researchers must identify IMU-derived metrics that are reliable and that provide similar estimates of gait outcomes to traditional motion capture. Furthermore, these metrics must be sensitive enough to capture age or knee OA-related differences. Therefore, the aim of this study was to calculate spatiotemporal and knee kinematic outcomes using IMUs, and to compare these outcomes to those calculated from simultaneously-captured motion capture data in young adults, older asymptomatic adults, and older adults with knee OA. Methods: Young (Y; age 21-35), older asymptomatic (Asymp; age 65-80, no history of chronic joint pain or diagnosis of OA), and older adults with knee OA (OA; age 65-80, Kellgren-Lawrence grade 2-4 in at least one knee) were recruited from the community using a University recruitment database, word of mouth, and via screening of medical records from a University orthopaedic clinic. The study included one visit, during which participants completed an IRB-approved consent form, a knee OA outcome score (KOOS) questionnaire, and overground gait testing. Participants were outfitted with four IMUs (Opal, APDM Inc.) and a standard 6 degree of freedom lower extremity marker set on the foot, shank, thigh, and pelvis for the right (Y and Asymp) or more symptomatic knee OA limb. IMU gyroscope and accelerometer axes were calibrated to world vertical, and to an approximation of each segment’s sagittal axes. Ten trials of overground gait were captured at 3 speeds: preferred and faster and slower than preferred. Walking speed, stride length, cadence, % gait cycle spent in stance, and sagittal plane knee range of motion were calculated via motion capture and IMU. All IMU variables except knee range of motion were calculated from the foot sensor using thresholds on accelerometer and gyroscope data to estimate when the foot was in contact with the ground and a continuous wavelet transform on accelerometer data to identify gait events. Knee range of motion was calculated using thigh and shank sensor gyroscope data. Motion capture variables were calculated via inverse kinematics (Visual3D, C-Motion). Outcome measures were compared between groups and testing devices (motion capture and IMU) for all walking speeds via MANOVA with α=0.05. Percent differences between methods were also calculated for descriptive purposes. Results: Current results include 6 individuals per group. Groups did not differ by height or body mass and Asymp and OA did not differ by age (Table 1). The overall MANOVA revealed differences between groups (p<0.01) but not devices (p=0.37). Average percent difference between methods for all groups and measures was 1.9% (range: 0.001-7.8%), with the largest difference seen in the slower speed condition for walking speed and stride length (% difference 3.9-7.8%). Between-group differences included speed and stride length at preferred walking speed, and knee range of motion at all walking speeds (all p<0.01, Table 2). Y walked faster and with longer stride lengths than both older groups, and Asymp walked faster and with longer stride lengths than OA. Knee range of motion was greater for Y and Asymp than OA at all walking speeds except faster than preferred, where OA and Y did not differ. Conclusions: Spatiotemporal gait measures and knee range of motion did not differ when assessed via standard optical motion capture or a simple IMU set. Moreover, IMU-derived values were as statistically accurate as standard motion capture even at slower speeds or for groups with potentially atypical gait dynamics (i.e., knee OA). Our results suggest that a minimal IMU setup can be used to accurately capture simple gait metrics and identify true differences in gait between different populations. This provides initial support for IMU measurement of gait outside of the laboratory setting.View Large Image Figure ViewerDownload Hi-res image Download (PPT)
Read full abstract
Nov 1, 2024
Miel Willems + 6
Just Published
Open Access