Suppressing Functional Electrical Stimulation Artifact from Facial Surface EMG Measurements

Methods for suppressing the electrical artifact that functional electrical stimulation introduces to surface electromyography (EMG) measurements are demonstrated. The methods are targeted for facial pacing for people who suffer from unilateral facial paralysis. The pacing includes the measurement of EMG signals from the healthy side of the face as a control signal to simultaneously activate the paralysed side with functional electrical stimulation.

Facial EMG signals typically have amplitudes up to a few hundreds of microvolts. The activation of facial muscles with functional electrical stimulation may require stimulation voltages that are more than 10^6 times larger. When electrical stimulation is fed to the paralysed one side of the face, the introduced voltage will also couple to the EMG measurements on the healthy side. This coupling is called the stimulation artifact. In the worst case, the EMG measurement inputs will saturate leaving it impossible to correctly detect facial muscle activations and their activation intensities. Additionally, the stimulation artifact may be erroneously detected as a muscle activation.

The presented methods for suppressing the stimulation artifact from EMG measurements include filtering implemented in the hardware and software, manipulating the stimulation waveform to help removing it with filtering, and sample-and-hold functionality implemented in the hardware of the EMG signal measurement chain to prevent amplifier saturation and to allow faster recovery from the artifacts. The methods are demonstrated with experimental results.