A Sequential Detection Method for Late Auditory Evoked Potentials

POTENTIALS U. Hoppe, U. Eysholdt Abteilung f ur Phoniatrie und P adaudiologie Universit ats{HNO{Klinik Friedrich{Alexander Universit at Erlangen, Germany S. Wei Signal Processing Division Dept. of Electronic and Electrical Eng. University of Strathclyde Glasgow, Scotland Abstract.{ This work presents a novel mechanism for detection of late auditory evoked potentials (AEP). AEPs, which are an important diagnostic tool to detect hearing de ciencies, are contained within the electroencephalogram (EEG) at a very low SNR. Our proposed automatic detection of AEPs is based on a Wavelet-Transform of EEG data for feature extraction. Several transform coe cients are then used for a classi cation by a neural network; its decisions on successive EEG segments are judged by a sequential statistical test. This test stops when a certain con dence is reached to either accept or reject an AEP. Besides providing an objective recognition of AEPs, this procedure can considerably reduce measurement time.