Spike detection using a syntactic pattern recognition approach

Syntactic methods were used to detect spikes utilizing time-domain features (slope and duration). The difference between this and earlier work is that here rules to discriminate each subject's spikes are automatically inferred from a training set consisting of sample spike waveforms obtained using expert knowledge, whereas the previous work used expert human knowledge to determine a set of universal rules. Therefore, the intersubject spike variability is accommodated in the design, without jeopardizing objectivity. The structural combination of the features is used by the syntactic method, providing a strict, explicit set of rules by which the spikes are identified. As a result, a rule-based definition for each subject's spike is obtained.<<ETX>>

[1]  F M Miezin,et al.  Computer quantification of electroencephalographic data recorded from renal patients. , 1975, Computers and biomedical research, an international journal.

[2]  P. Hazemann,et al.  Handbook of Electroencephalography and Clinical Neurophysiology , 1975 .

[3]  V. Radhakrishnan,et al.  Inference of regular grammars via skeletons , 1987, IEEE Transactions on Systems, Man, and Cybernetics.

[4]  J R Bourne,et al.  Artificial intelligence methods in quantitative electroencephalogram analysis. , 1982, Computer programs in biomedicine.