Multi-neural networks approaches for biomedical applications: classification of brainstem auditory evoked potentials

摘要

Auditory evoked potentials (AEPs) are electrical response caused by the brief stimulation of the auditory sensing system. AEP based techniques are important tools to diagnosis many of auditory pathologies. Especially to suspect the presence of auditory tumors called 'acoustic neuromas'. We investigate the design of a neural based biomedical diagnosis aide tool. We use three models of artificial neural networks: learning vector quantization, radial basis function and backpropagation ones. In our approach these three neural networks are used to achieve the classification in two multi-neural network configurations. A case study and experimental results are reported and discussed.