EIT Thigh Force Estimation

Abstract

Assessing muscle strength and estimating muscle force during everyday activities plays a crucial role in understanding human movement, rehabilitation, and sports science.
This project leverages Electrical Impedance Tomography (EIT), a non-invasive imaging technique that captures the internal conductivity distribution of tissues, to explore the feasibility of force level estimation from EIT data.

Inspired by EITPose, which demonstrated real-time monitoring of forearm muscle activity using EIT, we extend this approach to the thigh region.
A custom-built belt equipped with 16 electrodes was developed to record EIT data, while an Isoforce device simultaneously captured torque measurements.

The study is structured in two phases: - Estimation of discrete force levels (20–80 Nm) across multiple participants. - Continuous torque estimation, formulated as a regression problem.

Table of Contents

• Data Quality Check
• Data Preprocessing and Synchronization
• Dataset Generation
• PCA Analysis
• Classification
• Multi-Class Classification
• Regression
• Outlook
• Installation
• Author
• Acknowledgements

Data Acquisition

EIT signals and torque readings were captured using Sciospec and Isoforce systems. Example torque data from two different sources for assign the timestamps shown below:

Data Preprocessing and Synchronization

Raw measurements undergo filtering, alignment, and trial extraction before model training. Overview:

After acquiring the data it passed through the preprocessing pipeline before used for trainng data-driven models. The overview of this pipeline shown below:

Example (Participant 5): Filtered torque signal:

Classification Results

Two models were evaluated for multi-class force classification:

Regression Results

Continuous torque estimation with Random Forest:

Installation

Make sure you have Python 3.8 or later installed.
Clone the repository and install the required packages:

bash git clone https://github.com/Arash-Keshavarz/EIT_Thigh_Force_Estimation.git cd EIT_Thigh_Force_Estimation pip install -r requirements.txt

Author

This repository was created and is maintained by Arash Keshavarz, Institute of Communications Engineering, University of Rostock, Germany.

Contact: arashkeshavarzx@gmail.com

Acknowledgements

• The EITPose project served as an inspiration for extending EIT applications to the thigh region.
• Special thanks to the Institute of Communications Engineering, University of Rostock, for their support.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Contributing

Contributions are welcome! Please open issues or pull requests.