Signal Space Separation Algorithm and Its Application on Suppressing Artifacts Caused by Vagus Nerve Stimulation for Magnetoencephalography Recordings

Abstract

Magnetoencephalography (MEG) has been successfully applied to presurgical epilepsy foci localization and brain functional mapping. Because the neuronal magnetic signals from the brain are extremely weak, MEG measurement requires both low environment noise and the subject/patient being free of artifact-generating metal objects. This strict requirement makes it hard for patients with vagus nerve stimulator, or other similar medical devices, to benefit from the presurgical MEG examinations. Therefore, an approach that can effectively reduce the environmental noise and faithfully recover the brain signals is highly desirable. We applied spatiotemporal signal space separation method, an advanced signal processing approach that can recover bio-magnetic signal from inside the MEG sensor helmet and suppress external disturbance from outside the helmet in empirical MEG measurements, on MEG recordings from normal control subjects and patients who has vagus nerve stimulator. The original MEG recordings were heavily contaminated, and the data could not be assessed. After applying temporal signal space separation, the strong external artifacts from outside the brain were successfully removed, and the neuronal signal from the human brain was faithfully recovered. Both of the goodness-of-fit and 95% confident limit volume confirmed the significant improvement after temporal signal space separation. Hence, temporal signal space separation makes presurgical MEG examinations possible for patients with implanted vagus nerve stimulator or similar medical devices.