High-resolution time-frequency analysis of somatosensory evoked potential components by means of matching pursuit

This paper proposes to apply a high-resolution time-frequency analysis (TFA) algorithm, the matching pursuit (MP), to extract and identify detail components of somatosensory evoked potential (SEP) signals. Conventional TFA methods showed limited time-frequency resolution in short-period nonstationary SEP signals so that they cannot reveal detail components in time-frequency domain. The MP algorithm can decompose a SEP signal into a number of elementary components and provide a time-frequency parameter description of decomposed components. The stable components can be revealed by statistical analysis and classification of the extracted parameters. Experimental results on cortical SEP signals of rats show that a series of stable SEP components can be identified using the MP decomposition algorithm.