A Review of Major Methods of Evoked Response Analysis
Thalı́a Harmony · 2021
This chapter argues that application of a given procedure implicates precise assumptions, and it provides an overview of the stochastic properties of the ERs. The implication of this finding is that all procedures of analysis that assume the stationarity of the ERs, like Wiener filtering, adaptive filtering, amplitude sorting, Fourier transform, and so on, are severely handicapped if this assumption. Similarly, for application of linear discriminant analysis to different groups of averaged evoked responses, it is necessary to test the equality of the covariance matrices of the different groups. Considering a set of averaged evoked responses, principal component analysis may be performed in two basically different ways: Selection of the procedure should be made according to the goals pursued. Cluster analysis may be applied directly to the amplitude sampled ERs or to such transformations of these ERs as the Fourier transform or the results of principal component analysis.