We are delighted to share our latest paper on speech-induced vibration artifacts in intracranial recordings!
We set out to study basal ganglia-cortical interactions in speech production, acquiring electrocorticography and subcortical recordings during Deep Brain Stimulation (DBS) implantation surgeries in which participants performed a speech task. Many electrodes across different recording modalities and acquisition systems showed a signal that looked like the produced speech in frequency and timing and even had coherence with the recorded audio. Furthermore, we found the same signal in “blank” pins not connected to any electrode (but with wires that ran the extension of the acquisition chain) and on a vibration sensor attached to the stereotactic frame. Our data shows how speech-induced vibrations can create artifacts that affect the gamma band of intracranial recordings, the primary frequency range of interest for speech decoding. We are now working on methods to remove this artifact using machine-learning approaches!
This work wouldn’t have been possible without the generosity of all the fantastic participants who contributed time and effort to this study.
Comentários