Characterizing the Variability of Kinematic Outcome Measures and Compensatory Movements using Inertial Measurement Units

Abstract

Cost-effective wearable sensors to measure movement have gained traction as research and clinical tools. The potential to quantify movement with a portable and inexpensive way could provide benefits to patient populations (e.g. amputees) to supplement or replace current clinical evaluations. For example, characterization of frontal plane kinematic outcome measures is a relevant movement pattern to a complex amputee population. The ability to capture such movements could have important therapeutic opportunities. The current research worked towards characterizing frontal plane compensatory movement patterns with kinematic outcome measures described by inertial measurement units (IMU) data in healthy adults. This was an initial step towards developing a future toolkit that could characterize normal and aberrant movement patterns in clinical populations. The thesis is comprised of two related studies. The first study set out to evaluate the numerical accuracy of IMU estimated spatial measures when compared to a gold standard system. Six subjects completed six different movement tasks while instrumented with optical motion capture and IMUs. Each movement task probed the accuracy of specific deviations (e.g. vertical deviation). The hypothesis was that outcome measures would be strongly associated (r>0.8) and mean error would not be significantly different from zero and the coefficient of repeatability would be within priori set limits of agreement (±18 mm). Kinematic outcome measures had small mean error bias compared to gold standard measures and range of subject specific mean errors showed minimal differences. Task specific differences were evident when movement patterns exhibit large transverse rotations. These results showed the devices have a level of accuracy that may be suitable to characterize changes in movement patterns clinically. The second study aimed to utilize the same techniques from study 1 to describe compensatory kinematic outcome measures during a clinical obstacle avoidance task to differentiate between compensatory and normal movement patterns. Twelve subjects wore IMUs bilaterally on the ankles and on the belt above the right hip. An off the shelf orthotic knee brace was used to restrict lower limb knee joint kinematics (reduce range of motion). Participants completed 15 walking trials for three different brace conditions (No Brace, Unlocked Brace, Locked Brace) and two obstacle task conditions (Level Ground Walking and Obstacle Avoidance) to elicit a comparison of normal and compensatory movements. During the walking task, IMUs were able to characterize compensatory movements typical of the amputee population. Lateral deviation of the swinging foot was significantly larger during obstacle crossing with a locked brace compared to no brace. Maximum elevation of the limb was significantly larger while crossing obstacles compared to level ground walking and was precise enough to discern elevation differences of No Brace elevation from both Unlocked and Locked Brace conditions. Hip hiking was also significantly larger in the locked brace obstacle crossing from no brace obstacle crossing. Swing time was longer when the limb was braced and during obstacle crossing when compared to level ground walking. Healthy subjects had no significant changes to double support time compared those exhibited by amputees during walking. Overall, differences between IMU and gold standard measures are present. Mean error differences are present for certain tasks and criteria for agreeability between devices is not satisfied. Descriptive analysis of low subject mean error ranges across the majority of tasks indicate a potential utility in these measures to distinguish between movement patterns. During the clinical task, when knee mobility was manipulated compensatory movements were significantly different across conditions. This study provides evidence for the utility of IMU devices to support clinical gait analysis with quantifiable measures.