IMU-Based Gait Analysis

This repository contains scripts for IMU-based gait analysis:

• Calculating spatio-temporal gait paramters from raw IMU data
• Validating the gait parameters against reference systems using correlation analysis and Bland-Altman plots
• Visualizing the changes in gait patterns using foot movement trajectories and aggregated gait parameters
• Creating videos of synchromized video recording and sensor signals

Getting started

Getting the Project

git clone https://github.com/Linn39/imu_gait_analysis.git

Setting up the Environment

Once you have the project on your local machine, you can set up the Conda environment using the requirements.txt file.

conda create --name myenv --file requirements.txt python=3.9
conda activate myenv

Replace myenv with the name you want to give to your environment.

Alternatively, you can use pip to install the project requirements.

Calculate spatio-temporal gait parameters

• Get example data from Zenodo: StrokeGait or DUO-Gait, save the data in the project base directory. Remember to adapt the subject ID list according to the dataset.
• Run src/main_LFRF_preprocessing.py to prepare the IMU data.
• Run src/main_gait_parameters_charite.py to calculate the gait parameters.
• Modify experiment settings in src/LFRF_parameters/pipeline_playground.py if needed.

Example results

Manually inspecting the gyro threholds to recognize stance phases (when foot is stationary on the ground).

References

The algorithms are based on the following publications, with extended functionalities such as alternative algorithms, visualization of intermediate results for method development and debugging, and improved result plots.

• Trautmann, J., Zhou, L., Brahms, C. M., Tunca, C., Ersoy, C., Granacher, U., & Arnrich, B. “Tripod--A Treadmill Walking Dataset With IMU, Pressure-Distribution and Photoelectric Data for Gait Analysis”. In: Data 6.9 (2021), p. 95.
• Zhou, L., Fischer, E., Brahms, C. M., Granacher, U., & Arnrich, B. “DUO-Gait: A Gait Dataset for Walking Under Dual-Task and Fatigue Conditions with Inertial Measurement Units”. In: Scientific Data 10.543 (2023), pp. 1-10.
• Zhou, L., Rackoll, T., Ekrod, L., Balc, M. G., Klostermann, F., Arnrich, B., & Nave, A. H. “Monitoring and Visualizing Stroke Rehabilitation Progress Using Wearable Sensors” (In Press). In: 2024 46th Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC). 2024. Vol. 2024-July.
• Tunca, Can, et al. "Inertial sensor-based robust gait analysis in non-hospital settings for neurological disorders." Sensors 17.4 (2017): 825.

Validate gait quantification algorithms

• Get example data from Zenodo: TRIPOD. save the data in the project base directory.
• Run src/main_LFRF_preprocessing.py to prepare the IMU data.
• Run src/main_gait_parameters_TRIPOD.py to calculate the gait parameters and compare against reference systems.
• Modify experiment settings in src/LFRF_parameters/pipeline_playground.py if needed.

Data processing pipeline

Example results

References

The algorithms are based on the following publications, with extended functionalities such as alternative algorithms, visualization of intermediate results for method development and debugging, and improved result plots.

• Zhou, L., Tunca, C., Fischer, E., Brahms, C. M., Ersoy, C., Granacher, U., & Arnrich, B. “Validation of an IMU Gait Analysis Algorithm for Gait Monitoring in Daily Life Situations”. In: 2020 42nd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC). 2020, pp. 4229-4232.
• Zhou, L., Fischer, E., Tunca, C., Brahms, C. M., Ersoy, C., Granacher, U., & Arnrich, B. “How We Found Our IMU: Guidelines to IMU Selection and a Comparison of Seven IMUs for Pervasive Healthcare Applications”. In: Sensors 20.15 (2020), p. 4090.
• Trautmann, J., Zhou, L., Brahms, C. M., Tunca, C., Ersoy, C., Granacher, U., & Arnrich, B. “Tripod--A Treadmill Walking Dataset With IMU, Pressure-Distribution and Photoelectric Data for Gait Analysis”. In: Data 6.9 (2021), p. 95.

Quantifying and visualizing gait changes

• Get example data from Zenodo: StrokeGait, save the data in the project base directory.
• Run src/main_LFRF_preprocessing.py to prepare the IMU data.
• Run src/main_gait_parameters_charite.py to calculate the gait parameters.
• Run main_plot_aggregated_trajectories.py to visualize the foot movement trajectories.
• Run main_plot_gait_parameters.py to visualize the foot movement trajectories.

Example results

References

The algorithms are based on the following publication, with extended functionalities such as visualization of intermediate results for method development and debugging, and improved result plots.

• Zhou, L., Rackoll, T., Ekrod, L., Balc, M. G., Klostermann, F., Arnrich, B., & Nave, A. H. “Monitoring and Visualizing Stroke Rehabilitation Progress Using Wearable Sensors” (In Press). In: 2024 46th Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC). 2024. Vol. 2024-July.

Create sensor videos

This is used for presentations/demos.

• Trim the original video recording, so that it starts with the first initial contact of the right foot and has the deseired length. Save the trimmed video into ./data/processed/sensor_video/video_trim_{sub}_{visit}.mp4
• Run ./src/main_sensor_videos.py to generate the IMU videos, which are saved in ./data/processed/sensor_video/
• Check the initial contact time for the IMU signal from the right foot, update config["imu_ic"]
• Run ./src/main_sensor_videos.py again (comment out the part that generates the IMU videos, since the videos were already created in previous steps) to generate the merged video, which are saved in ./data/processed/sensor_video/
• Trim the final merged video if necessary

Example results

Video recording with simultaneous raw IMU signals from the right wrist (RW), right foot (RF), and lower back (ST).

Project contributers

• Lin Zhou
• Justin Trautmann
• Eric Fischer
• Can Tunca