Open Access Dataset, Toolbox and Benchmark Processing Results of High-Density Surface Electromyogram Recordings.

Area of Science:

• Biomedical Engineering
• Neuroscience
• Rehabilitation Engineering

Background:

• High-density Surface Electromyogram (HD-sEMG) recordings are crucial for understanding muscle activity.
• Existing datasets may lack the comprehensive nature required for advanced neural interface research.
• Dexterous finger manipulations present complex electromyographical signals.

Purpose of the Study:

• To introduce the Hyser dataset, a comprehensive collection of HD-sEMG recordings.
• To provide a versatile toolbox for analyzing HD-sEMG data and performing signal decomposition.
• To offer benchmark results for pattern recognition and EMG-force applications.

Main Methods:

• Acquired 256-channel HD-sEMG data from 20 subjects across two sessions.
• Collected data during various tasks including hand gestures, maximal voluntary contractions, and multi-finger force control.
• Developed a toolbox for dataset analysis and motor unit decomposition using independent component analysis.

Main Results:

• The Hyser dataset comprises five distinct sub-datasets tailored for gesture recognition and proportional force control.
• Benchmark results for pattern recognition and EMG-force applications are provided.
• The toolbox enables standard analysis methods and advanced HD-sEMG signal decomposition.

Conclusions:

• The Hyser dataset and toolbox offer a valuable, open-access resource for the neural interface research community.
• This resource facilitates research in gesture recognition, neuroprosthetics control, and neural rehabilitation.
• The provided benchmark analyses promote collaboration and accelerate advancements in the field.