Determination of the reliability of a gait analysis method based upon motion sensors and a data processing software throughout a comparative analysis using a photogrammetry system.

Abstract

Introduction 

The use of motion sensors for gait analysis offers an alternative to photogrammetry techniques, provided that the analysis method designed by the researcher is effective. Sensors can be an effective tool for diagnosis, treatment planning, and patient counselling, as well as identifying key events such as foot strike (FS) and toe-off (TO). However, their effectiveness depends on the tool used to acquire and process the signals. A gait analysis method based on inertial sensors and processing software was developed to determine whether it is reliable for detecting key events. One way to assess the reliability of this methodology is by estimating the error in detecting gait events measured with sensors and comparing it to those obtained via photogrammetry with the means of determining whether they can be used as a tool for motion analysis.

Objective

Evaluate the performance of a gait analysis method based on inertial sensors (IMUs) and processing software, by comparing it with a photogrammetry system, in order to determine its reliability in the measurement and analysis of gait for patient evaluation and treatment.

Methodology

Gait measurements were taken using Xsens inertial sensors alongside the gold standard Vicon Nexus system, placing markers according to the PlugIn Gait Lower Body model. A total of 7 IMUs were used: two were placed on the thighs, two on the right and left shanks, two on each foot, and one at the level of the second sacral vertebra. Measurements were taken from 20 healthy individuals—16 men with an average age of 23.9 (2.05) years and 4 women with an average age of 24.9 (2.95) years. The data obtained were processed in MATLAB to detect FS and TO events. The program allows for comparison of the detection times between systems and calculates error to determine the reliability of the sensors.

Results 

When calculating the error based on the time difference of 858 events detected by Vicon versus Xsens, a mean error of 15.16 ms was obtained, with a standard deviation of 106.66 ms, an RMS of 107.67 ms, and an effect size of 0.10. This indicates that using IMUs with the proposed methodology has the same reliability for detecting key events as the photogrammetry system.

Conclusion

Using IMUs with the proposed methodology is just as reliable as the gold standard for detecting temporal variables, as evidenced by an effect size of 0.10—representing the mean difference between groups and the weighted standard deviation. It is also concluded that inertial sensors can be an accessible alternative for identifying key gait events and, from these, obtaining spatiotemporal gait parameters.