Interpretation of the Biomedical Signals using the RBF-Type Neural Networks

Abstract:-The paper described determination of parameters of Brainstem Evoked Auditory Potentials (BAEP) registration using computational intelligence methods. In the research on the system of automated diagnosis of the patient's auditory system the Radial Basis Functions (RBF) neural networks have been applied. The authors have also presented an interpretation of the determined network's weight coefficients, which leads to an easy evaluation of the required characteristics of the BAEP registration. Diagnostic systems based in the analysis of BAEP signals are currently a dynamically developing field of objective research concerning the human auditory system. The physician's interpretation usually includes only the amplitudes and latency periods of characteristic waves, the presence and time location of which, provides a basis for diagnosing the patient's hearing abilities. Because of great diversity of signals, which in general depend on the age, sex and also on the stimulus intensity, the evaluation and elaboration of a unanimous diagnosis becomes quite difficult for inexperienced physicians. Therefore the authors of the present paper make an attempt of constructing a diagnostic system, which could be able to eliminate the uncertainty element from the process of evaluation of hearing abilities, and would be able to provide a fully autonomous operation, based on clear, predefined criteria