EMG-based characterization of pathological tremor using the iterated Hilbert transform

Area of Science:

• Biomedical Engineering
• Neuroscience
• Signal Processing

Background:

• Pathological tremor characterization is crucial for developing effective tremor suppression techniques, such as functional electrical stimulation.
• Accurate estimation of tremor's amplitude, phase, and frequency from muscle activity (EMG) or movement (kinematics) is essential.
• Existing methods may lack accuracy or require prior knowledge of tremor characteristics.

Purpose of the Study:

• To present and validate a novel approach for estimating pathological tremor characteristics from surface electromyogram (EMG) signals.
• To demonstrate the efficacy of the iterated Hilbert transform (IHT) in modeling tremor and voluntary activity components.
• To assess the performance of the IHT method against simulated and experimental pathological tremor data.

Main Methods:

• Utilized the iterated Hilbert transform (IHT) for signal processing of surface electromyogram (EMG) data.
• Tested the IHT method on simulated tremor signals from a tremor generation model and experimentally recorded patient data.
• Compared the IHT method's performance with empirical mode decomposition (EMD) for tremor characterization.

Main Results:

• The IHT method effectively demodulated tremor amplitude (R²=0.52), frequency (RMSE=2.6 Hz), and phase.
• Accurate estimation of voluntary activity was achieved (R²=0.62 with simulated inertial load).
• Estimated tremor components from experimental data showed high correlation with inertial measurements of limb movement (0.62±0.15).

Conclusions:

• The iterated Hilbert transform (IHT) provides an accurate and effective method for characterizing pathological tremor from EMG signals.
• The IHT method outperforms empirical mode decomposition in tremor analysis without requiring a priori knowledge.
• This approach holds potential for integration into control systems for tremor suppression strategies.