Visual evoked potential classification for clinical use

Visual evoked potential (VEP) classification is very useful for better VEP exploitation in clinical practice. A discrimination system has been proposed for VEP selective averaging and diagnostics. The system has a tree-like architecture allowing various combination possibilities. The principal components analysis is used as main modeling method, while VEP classification is essentially performed by multilayer networks. The final decision is made by the synthesis unit of the system. The system is developed for real-time processing. The experimental results have shown that estimated VEP is effectively enhanced by selecting trails of better "quality", and the system is capable to distinguish VEPs in response to different stimuli.