Date of Award
Master of Science (MS)
Hutchison , Randy
DesJardins , John
Evaluating the human gait cycle with inertial measurement units (IMU) may prove beneficial for applications such as diagnoses of musculoskeletal diseases and assessment of rehabilitation regimes. An IMU system is potentially applicable for diagnosing and assessing rehabilitation outcomes for a variety of neuromuscular diseases since it is small, portable, and less expensive than a camera system. IMUs directly measure angular velocity, whereas position data from a camera system must be processed twice to obtain this information. The purpose of this research is to determine repeatability of IMU angular velocity data, and agreement between angular velocity data from an IMU system and a camera system during normal gait. From this data, the feasibility of using IMU systems in clinical or rehabilitative settings for obtaining reliable angular velocity data will be determined.
Lower limb motion data was collected simultaneously from six XSens MTx IMUs (XSens Technologies, Enschede, The Netherlands) and an 8-camera Qualisys Motion Capture system (Pro-Reflex, 240 Hz system). Each IMU consists of three orthogonal accelerometers, gyroscopes, and magnetometers. Data from 4 subjects (3 males, 2 females) were collected after an initialization technique before each trial to reduce effects of electro-magnetic interference with the IMUs. Knee joint angular velocities (Gx, Gy, Gz) corresponding to appropriate knee joint angles (flexion/extension, adduction/abduction, and internal/external rotations) from both systems were used in this analysis. Coefficients of variation (COV) were calculated for both IMU and camera data to determine variability of data from both systems. Knee joint Average angular velocities from both systems for each subject and limb were plotted together to visually evaluate correlation between data sets. F-test analyses were performed on linear models of the data to determine areas of co-linearity within the gait cycle, and at different intervals of angular velocities.
The IMUs had lower COV's than the camera system, likely due to the fact that the IMUs directly measure angular velocity, and camera system derives angular velocity from position data. However, these differences were not statistically different, likely due to variability within trials for individual subjects. Linearity between camera system and IMU angular velocity was visually observed only about the flexion/extension axis during segments of the gait cycle occurring from 0-4% (heel strike) and 65-100% (swing phase) of the gait cycle. Comparisons about the adduction/abduction and internal/external axes showed evidence of linearity for lower angular velocities. Linear regression statistics showed that the only correlational trend between the two systems was around 8-12% of the gait cycle for all three rotational axes. This may be due to drift of the IMU data. Although the camera system is the 'gold standard' in motion analysis, IMUs may be used for applications in which angular velocity for a flexion-extension movement at low joint angles is being evaluated. Future studies will include a larger sample population, and evaluate specific movements within human gait that affect drift of the IMUs. In addition, other IMU system designs could be evaluated for clinical use, and other algorithms that further reduce the effects of drift should be implemented.
Ex-lubeskie, Chelsea, "Evaluation of Angular Velocity Data from Inertial Measurement Units for Use in Clinical Settings" (2013). All Theses. 1598.