Temporal Gait Analysis

Temporal spatial gait analysis is commonly used to quantify gait function in healthy and pathologic populations. Collecting complete temporal or spatial gait assessments often required instrumentation. The key to any instrumentation used to calculate temporal gait variables is to accurately identify first and terminal contact events of a step. These events are called heel strike and toe off for typical gait patterns. The identification of these gait events allows for calculation of variables such as stride time, stance time, swing time and cadence.

The Use of IMUs

There is considerable interest in using wearable sensors such as IMUs to capture temporal gait data inside and outside the laboratory setting. IMUs allow for flexibility in protocol design as they are not limited to a confined space. However, with this flexibility comes the need for validation and optimization. IMUs must be placed on the body of subjects in specific locations and algorithms are used to interpret the linear and angular acceleration signals to identify key gait events. In the case of a temporal gait analysis, the algorithms use IMU data output to calculate heel strike and toe off. Algorithms are often developed to work with specific sensor sets placed in specific locations on the body. They are also developed for specific tasks. Recent research from the FDA looks to determine how to best optimize IMU-based systems for the use in temporal gait analysis

Gaps in research

To date, most algorithms for gait event detection have been developed for straight line walking. This is problematic when the end goal is to use the IMU systems to capture real-world data. Real-world walking involves a mixture of straight line walking, turning and curved path walking. There are currently gaps in the research to determine if the algorithms developed with straight line walking are accurate for real-world data collection.

The use of IMUs for gait analysis has little agreement on sensor placement, as sensors are often placed on the lower back, shank and/or feet. Algorithms need to be written for specific sensor placement locations, using different data analysis techniques to identify key gait events. A recent published literature review found 17 different algorithms for identifying key gait events. Research is necessary to ideal sensor placement and what data analysis techniques will optimize the accuracy of these systems.

The goal of this study is to provide guidance for the use of wearable sensor systems for both real-world and laboratory-based temporal gait analysis.

Study Methods

IMU placements were chosen based on previous literature, using 5 sensors placed at on the shank and foot of the subjects. Previous literature has showed shank and foot sensor placement has been more accurate than placement on the lower back. Walking tasks were completed on the Zeno Walkway System and used PKMAS to determine the initial and terminal contact events to be used as reference points concurrently with IMU system analysis. Straight line and obstacle avoidance gait tasks were performed, which led to both straight line and curved path walking taking place.

5 different algorithms were used to compute the initial and terminal contact events from the IMU sensor data. Those event times were compared to the reference events from the Zeno Walkway System to determine the accuracy of the algorithms.  The algorithms each used different, commonly used techniques to calculate gait events. Gait events in each system were used to calculate stance time, swing time and stride time.

Study Implications

This study set out to better understand the error from clinical measurement devices that use IMUs to calculate temporal gait metrics and provide insights to minimize those errors. The resulting errors seen from different sensor placements and different algorithms can provide guidance. For a system to have clinical utility, the error from the device must be lower than the minimal detectable difference in what is being studied.

Errors from IMU-based gait analysis systems can come from the location in which the sensor(s) is/are placed, the algorithm used, and the testing protocol. In order to maximize the accuracy of a wearable system for temporal gait analysis, the sensor location, algorithm and testing protocol must be validated and optimized, ideally, comparing the IMU system to a gold standard measurement device. To read complete results from this study, the publication is located here: https://pubmed.ncbi.nlm.nih.gov/33105876/