Computer analysis of auditory brainstem evoked potentials

摘要

The paper presents issues associated with examination of auditory evoked potentials. Considering that the current methods are mainly based on timing interpretation of waveform potentials that requires significant experience from examining neurologist, the authors have undertaken the task aimed at elimination of subjective assessment of results and automation of the tests. By applying methods of digital signal processing, the authors have isolated distinctive features of evoked potentials, which allows for differentiating the normal results from pathological cases. The use of wavelet decomposition and linear SVM network has ensured correct classification of 40 different clinical cases, including 20 normal and 20 pathological cases.