Knowledge-based systems for neuroelectric signal processing
A.S. Sehmi, N.B. Jones, S.Q. Wang, Gareth H. Loudon · IEE Proceedings - Science Measurement and Technology · 1994
This paper describes expert systems suitable for signal processing and decision support in the interpretation of neuroelectrical signals such as brainstem auditory evoked potentials (BAEP), interference pattern electromyograms (EMG) and electroencephalograms (EEG). These systems are characterised by a significant amount of coupling between numerical and symbolic processing techniques. The BAEP and EMG expert systems incorporate rule-based inference mechanisms with a high degree of uncertain inference using fuzzy logic. The EEG expert system uses an object-oriented approach to capture high-level stereotypes of spatiotemporal concepts in multichannel EEG signals. These stereotypes can trigger lower-level numerical procedures in an opportunistic manner to extract contextual numerical information using a limited form of uncertain inference. A conceptual hardware and software framework for implementing such expert systems is also outlined.