Detection and multichannel SVD-based filtering of trigeminal somatosensory evoked potentials

Very weak and noisy trigeminal somatosensory evoked potentials (TSEPs) are considered, which are successfully evoked by electrical stimulation of the trigeminal nerve of 15 patients with endosseous oral implants. As TSEP analysis provides an objective means of assessing neuronal function, it is considered to be a promising tool for investigating tactile sensation through anchoring implants in bone. For this purpose, a study of TSEP signals acquired from patients with endosseous oral implants has been carried out. Since TSEPs are severely contaminated by background ongoing electrical activities of the brain, a methodology is developed for statistically detecting the transient signal (TSEP) in the biological noise (EEG). For nine out of 15 patients, transient signals are defected in the background EEG activity. The TSEPs of these nine patients are subjected to further analysis. A multichannel singular value decomposition (SVD)-based filtering method is applied which successfully separates out the most energetic TSEPs from the background EEG, thereby increasing significantly the SNR of the recorded signals and improving extraction of the characteristic components of the TSEPs. It is shown that the most prominent feature of the TSEP signals for patients with endosseous oral implants is a wave with peak latency between 9 and 15 ms, generally followed by a wave between 25 and 28 ms or 34 and 38 ms for the specific cortical response areas.