Promediación coherente mejorada mediante transformada wavelet de potenciales evocados auditivos de tronco cerebral

Abstract

The brainstem auditory evoked potentials (BAEP) are immersed in physiological and non physiological activity. To estimate these potentials the coherent average is the most utilized and accepted technique. Nevertheless, it supposes that the evoked potentials signal is stable in each epoch, which is not certain in most cases and consequently the signal is distorted. Besides, the time required to obtain a reliable estimation of the signal and its reply is very long. For these reasons, it becomes necessary to find different alternatives to estimate the BAEP signal. In this work, a discrete wavelet transform is applied to improve the signal-to-noise ratio of the BAEP, looking for a considerable reduction in the estimation time required with the classical coherent average method. Through multiresolution analysis, the optimum coefficient's distibution for an accurate BAEP reconstruction was determined. Finally, a denoising process on averages of 200, 400, 600, 800 and 1000 epochs was implemented to determine the minimum number of epochs required for a reliable signal estimation. The results show that 800 epochs are enough to obtain equivalent results to the average of 2000 epochs. © Springer-Verlag Berlin Heidelberg 2007.