Lower arm electromyography (EMG) activity detection using local binary patterns.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Human-Computer Interaction

Background:

• Myoelectric signals are crucial for understanding muscle activity.
• Accurate detection of muscle activity is essential for prosthetic control and human-computer interfaces.
• Existing activity detection methods often require complex tuning and are sensitive to noise.

Purpose of the Study:

• To introduce a novel electromyography (EMG) activity detection technique using 1-D local binary pattern histograms.
• To simplify the implementation and improve the robustness of EMG activity detection.
• To evaluate the performance of the new method against existing techniques.

Main Methods:

• Application of 1-D local binary pattern histograms to forearm surface myoelectric signals.
• Multi-channel activity detection with minimal parameters and no majority vote.
• Elimination of per-channel threshold tuning.
• No requirement for pre-acquisition of quiescent signal properties.

Main Results:

• The proposed method demonstrated effective discrimination between muscle activity and inactivity.
• Activity detection was performed across multiple channels efficiently.
• The algorithm showed superior performance compared to offline single- and double-threshold methods.
• Enhanced tolerance to noise was observed in real-world datasets.

Conclusions:

• The 1-D local binary pattern histogram technique offers a simplified and robust approach to EMG activity detection.
• This method reduces implementation complexity and error proneness.
• The technique shows significant potential for applications requiring reliable myoelectric signal analysis, such as prosthetic limb control.