Evoked potential signal analysis in brain injury detection
MetadataShow full item record
A study of relations between changes in evoked potential (EP) signal and possible brain injuries is reported in this thesis. Monitoring of EP signals proves to be very useful in neuro-critical care units and highrisk operations. Research results reported here may result in practical neuro-electric signal monitoring instruments. Advanced signal processing algorithms are used to enhance and analyze the EP signals and to quantify changes in the EP signals. This thesis illustrates the analysis of various signal processing algorithms in tracking the EP variations. The Fourier Linear Combiner (FLC) and the Adaptive Itakura Distance measure (AID) are investigated for tracking EP morphological variations. Data analyses show that FLC is better suited as a real-time method of tracking the morphological changes. Comparative analysis of the Generalized Cross-Correlation method (GCC), the Time Delay Estimation via Least Mean Square algorithm (LMSTDE), and the Adaptive Time Delay Estimation algorithm (ATDE) are carried out to evaluate their effectiveness of reliably estimating latency variations. Enhancements to the algorithms have been made so that they are better suited to this particular application. The following issues are addressed: 1) the accuracy in latency estimation, 2) robustness to the background electroencephalograph (EEG) signals and to the simultaneously occurring morphological changes, 3) feasibility and effectiveness of the algorithm as a real-time latency estimation method. The potential usefulness of the ATDE in this area, compared to the other algorithms, is established via simulation studies. Analyses of actual EP recordings from an impact acceleration experiment are conducted for detecting possible injuries.