IntellEvent

Introduction

IntellEvent is a robust deep learning-based framework for gait event detection across various pathologies for 3D motion capture data. By leveraging deep learning models, IntellEvent accurately detects gait events (initial contact (IC) and foot off (FO)) in patients with different clinical conditions, including malrotation deformities and/or frontal malalignments of the lower extremities, club foot, cerebral palsy, drop foot, and healthy participants. IntellEvent ensures reliable and precise gait events even in complex pathological cases (IC: < 5.5 ms @150 Hz, FO: < 11.4 ms @150 Hz ). For more detailed information, refer to the original paper: Robust deep learning-based gait event detection across various pathologies.

Dataset

The dataset used for IntellEvent consists of a comprehensive retrospective clinical 3D gait analysis (3DGA) dataset:

• Total Subjects: 1211 patients and 61 healthy controls

Categories

• Malrotation deformities of the lower limbs (MD): 730 subjects
• Club foot (CF): 120 subjects
• Cerebral palsy (CP): 344 subjects
• Cerebral palsy with only drop foot characteristics (DF): 17 subjects
• Healthy controls (HC): 61 subjects

Requirements

This framework has been tested with Vicon Nexus version 2.14 and higher. No installation require!
If you would like to use Vicon Nexus 2.12.1, please get in touch, we will find a solution.

Vicon Nexus Usage

1. Download the 25_IntellEvent.zip folder from the release here.
2. Extract the files to a folder of your choice. Note: All files must be loacated in the same folder.
   
   
   
   
3. Start the vicon_server.exe.
   
   
   
   
4. Create a new Run Python Operation in a Vicon Nexus pipeline from the operation Data Processing tab. Add the vicon_pipeline.py to the Python script file path.
   
   
   
   
5. Run the pipeline and save time!

Future Developments

• Integrating Further Movement Tasks:
  • Turning
  • Running
• Ensuring Robustness for Different Laboratory Settings:
  • Standardize data preprocessing from multiple laboratory sources
  • Utilize data from different labs
• Integrating Fine-Tuning Pipeline
• Implement a Pipeline for Continuous Refinement and Optimization

Current Results

The current model achieves the following Mean Absolute Errors (MAE) in milliseconds for different pathologies:

Category	MD	CF	DF	CP	HC
IC MAE [ms]	2.7	3.5	5.4	4.9	2.5
FO MAE [ms]	7.9	8.7	9.9	11.3	8.3

Citation

If you are using IntellEvent in your research we would appreciate a citation.

[1] B. Dumphart et al., ‘Robust deep learning-based gait event detection across various pathologies’, PLOS ONE, vol. 18, no. 8, p. e0288555, Aug. 2023, doi: 10.1371/journal.pone.0288555.

@article{dumphartRobustDeepLearningbased2023,
  title = {Robust Deep Learning-Based Gait Event Detection across Various Pathologies},
  author = {Dumphart, Bernhard and Slijepcevic, Djordje and Zeppelzauer, Matthias and Kranzl, Andreas and Unglaube, Fabian and Baca, Arnold and Horsak, Brian},
  year = {2023},
  journal = {PLOS ONE},
  volume = {18},
  number = {8},
  pages = {e0288555},
  publisher = {{Public Library of Science}},
  issn = {1932-6203},
  doi = {10.1371/journal.pone.0288555},
  keywords = {Algorithms,Cerebral palsy,Feet,Gait analysis,Machine learning algorithms,Neural networks,Recurrent neural networks,Toes}
}

Contact

If you need any help, have further ideas, or have questions regarding IntellEvent please feel free to contact me!

[email protected]

[email protected]

[email protected]

License

Creative Commons Attribution 4.0 International Public License