Adaptive Walsh Estimation Of Time-varying Evoked Potential Signal
Xinrong Guo, Nitish V. Thakor · 2005
This paper presents a new method, adaptive Walsh Function Model (WFM), to estimate time-varying Evoked Potential (EP) signals. Using real EP signals plus real noise as inputs to the model, we compare the performance of two models: WFM and Fourier Series Model (FSM). Both models are also used to trace EP signal during experimental and clinical procedures. The results show: 1) Both WFM and FSM perform satisfactorily in detecting and tracing time-varying EP signals in high levels of noise; 2) WFM needs less memory and calculations than FSM; 3) For the same adaptive step size p I WFM traces faster transients than FSM but WFM attains lower steady-state signal to noise ratio (SNR) ; 4) The WFM and FSM are suitable not only for the detection of EP signals but other periodic or quasiperiodic signals as well.