Related Experiment Video
Updated: Dec 31, 2025
08:15
Capturing Dynamic Finger Gesturing with High-resolution Surface Electromyography and Computer Vision
Published on: March 28, 2025
1.1K
Investigation of Channel Selection for Gesture Classification for Prosthesis Control Using Force Myography: A Case
Chakaveh Ahmadizadeh1, Brittany Pousett2, Carlo Menon1
1Menrva Research Group, Schools of Mechatronic Systems Engineering and Engineering Science, Simon Fraser University, Metro Vancouver, BC, Canada.
Frontiers in Bioengineering and Biotechnology
|January 11, 2020
Summary
This study introduces a design method for prosthetic control using fewer sensors, achieving performance comparable to high-density systems. Channel selection methods significantly reduce sensor count while maintaining accuracy for advanced prosthetic devices.
Area of Science:
- Biomedical Engineering
- Rehabilitation Engineering
- Human-Machine Interfaces
Background:
- Force Myography (FMG) is a promising human-machine interface (HMI) for controlling prosthetic devices.
- High-density FMG (HD-FMG) offers higher accuracy but increases system complexity and cost.
- A need exists for simpler, cost-effective FMG systems without compromising performance.
Purpose of the Study:
- To propose and assess a design method for powered prostheses using fewer sensors.
- To determine the efficacy of various channel selection (CS) methods for optimizing sensor placement.
- To achieve performance comparable to HD-FMG systems with a reduced sensor count.
Main Methods:
- Utilized three datasets from force sensitive resistors in a transradial amputee's prosthesis socket.
- Collected sensor data during six distinct gestures performed five times each.
- Evaluated five CS methods: Sequential Forward Selection (SFS), Minimum Redundancy-Maximum Relevance (mRMR), Genetic Algorithm (GA), and Boruta.
Main Results:
- Three CS methods (mRMR, GA, Boruta) significantly reduced sensor channels while preserving classification accuracy across all datasets.
- The Genetic Algorithm (GA) consistently yielded the smallest channel subsets.
- Boruta and mRMR demonstrated greater stability (consistency) compared to GA when data varied.
Conclusions:
- The proposed design method is feasible for creating simpler, high-performing prosthetic systems.
- Channel selection effectively reduces sensor requirements for FMG-controlled prostheses.
- This approach offers a practical solution for developing advanced, yet cost-effective, prosthetic devices.
