Detection, Multichannel SVD Based Filtering and Characterization of Trigeminal Somatosensory Evoked Potentials

The very weak and noisy trigeminal somatosensory evoked potentials (TSEPs) are considered, which are evoked by electrical stimulation of an endosseous oral implant. The aim is to make the tactile sensation through endosseous implants objective. Although a lot of nerves are damaged by extraction of teeth, it has been reported from psychophysical tests [11] that a patient rehabilitated with an osseointegrated implant seems to have a subjectively improved ability to feel through his prosthesis and the anchoring implant in the bone. We want to find an objective and non-invasive measuring method for the investigation of the trigeminal nerve and try to explain this feedback mechanism. The TSEPs will be applied for this purpose, but these are very noisy and weak signals. Subsequently, a methodology is designed for statistically detecting the transient signal (TSEP) in the biological noise and a multichannel SVD (singular value decomposition) based filtering method is described in order to improve the signal-to-noise ratio of the recorded signals and to estimate the very diffuse TSEP waveform. Furthermore, as this filtering has the ability to investigate several channels simultaneously, the correlation between the channels can be exploited, which is appropriate for the determination of the characteristic components of the TSEPs, which are the peak latencies and amplitudes. Keywords— evoked potential, multichannel SVD based filtering, correlation coefficient, peak identification