BioPatRec: A modular research platform for the control of artificial limbs based on pattern recognition algorithms

Area of Science:

• Biomedical Engineering
• Computer Science
• Rehabilitation Engineering

Background:

• Myoelectric signal processing and pattern recognition are crucial for prosthetic control.
• Inter-study comparisons are hindered by study-dependent variables.
• A common research platform is needed for algorithm development and evaluation.

Purpose of the Study:

• Introduce BioPatRec as an open-source software platform.
• Provide a standardized environment for developing and evaluating prosthetic control algorithms.
• Facilitate seamless implementation of signal processing, feature selection, pattern recognition, and real-time control.

Main Methods:

• BioPatRec offers modular and customizable functionalities for various algorithms.
• It supports seamless benchmarking of algorithms from different research fields.
• Includes implementation of signal processing, feature extraction, pattern recognition, and real-time control.

Main Results:

• Demonstrated BioPatRec with Regulatory Feedback Networks (RFN), achieving comparable results to established classifiers.
• Released with Linear Discriminant Analysis and Multi-Layer Perceptron classifiers.
• Provides routines for myoelectric control of a virtual hand, from data acquisition to real-time evaluation.
• Includes a publicly available repository of myoelectric signals for algorithm benchmarking.

Conclusions:

• BioPatRec is openly available to accelerate algorithm development for improved prosthetic control.
• Facilitates community contributions and collaboration across disciplines and continents.
• Aims to enhance the quality of life for patients through advanced prosthetic technology.