Computational Intelligence in Electroencephalogram Analysis - Computational Intelligence in Biomedical Engineering - page 280

EEG signal rearrangement

Feature extraction

Feature selection and reduction

Unsupervised fuzzy clustering

Dynamic state recognition and identification

Epileptic events detection and forecasting
by the dynamic state identification

FIGURE 6.6
Block diagram of the fuzzy clustering technique used by Geva and Kerem
(1998) to identify epileptic events using the EEG waveforms. Features were
extracted using techniques such as statistical moments calculations, correla-
tion analysis, spectrum estimation, and time-frequency decomposition. Fea-
tures were reduced using principal component analysis (PCA). (Adapted from
Geva, A. B. and Kerem, D. H., IEEE Trans. Biomed. Eng., 45, 1205-1216,
1998. c IEEE.)

have a sensitivity of 62.5% with 90.5% specificity. The authors highlighted the
importance of using individualized data for developing classifiers for epileptic
seizure prediction.

6.2.1.2 Huntington's Disease

HD is a genetically inherited neurological condition, which affects a signifi-
cant proportion of the population. The disease is caused by a constant expan-
sion in the Huntingtin gene that leads to cell death in regions of the brain