Digital computer analysis of the human EEG in psychiatric research

An ensemble system for automatic sleep stage classification using single channel EEG signal

The present work aims at automatic identification of various sleep stages like, sleep stages 1, 2, slow wave sleep (sleep stages 3 and 4), REM sleep and wakefulness from single channel EEG signal. Automatic scoring of sleep stages was performed with the help of pattern recognition technique which involves feature extraction, selection and finally classification. Total 39 numbers of features from time domain, frequency domain and from non-linear analysis were extracted. After extraction of features, SVM based recursive feature elimination (RFE) technique was used to find the optimum number of feature subset which can provide significant classification performance with reduced number of features for the five different sleep stages. Finally for classification, binary SVMs were combined with one-against-all (OAA) strategy. Careful extraction and selection of optimum feature subset helped to reduce the classification error to 8.9% for training dataset, validated by k-fold cross-validation (CV) technique and 10.61% in the case of independent testing dataset. Agreement of the estimated sleep stages with those obtained by expert scoring for all sleep stages of training dataset was 0.877 and for independent testing dataset it was 0.8572. The proposed ensemble SVM-based method could be used as an efficient and cost-effective method for sleep staging with the advantage of reducing stress and burden imposed on subjects.

The discovery of antidepressant drugs by computer-analyzed human cerebral bio-electrical potentials (CEEG)

Antidepressant properties of six compounds were predicted based on their computer-analyzed human electroencephalographical (CEEG) profiles. The clinical investigations with mianserin (GB-94) confirmed the CEEG prediction. This compound has now been marketed as the first antidepressant of which the clinical effects were discovered solely by the quantitative pharmaco-EEG method. As predicted by the CEEG, clinical antidepressant properties of GC-46, mesterolone, and estradiol valerate were observed in preliminary investigations. No extensive studies with definite statistical results were yet carried out with these compounds. No systematic large studies could be conducted with cyclozocine and cyproterone acetate because of the intolerable side effects with these compounds. The optical isomers of mianserin, GF-59 and GF-60, both predicted as antidepressant by the computer EEG data base, have not yet been tested in depressive patients. None of these compounds possess the "typical" pharmacological and/or biochemical profiles of marketed antidepressants. Thus, the discovery of the established antidepressant properties of mianserin (GB-94) by computer analyzed EEG method challenges the well-known biochemical hypotheses of depression and the "classical" development of antidepressant drugs.

Serial quantitative electroencephalography

Repeated EEG examinations are often used in diagnostically difficult cases to assess whether an EEG abnormality is progressive, regressive, or stationary. The conventional visual method is, however, poorly adapted to consider the variability when consecutive EEG recordings are compared. Using computer analysis of EEGs, the feature of the recordings as well as their variances can be measured, and the differences between the recordings can be tested statistically. The aim of the present study was to develop a practicable method for quantitative serial EEG examinations. Twenty EEGs from 5 volunteers were recorded to obtain reference data. The data acquisition was automated to obtain data without visual bias. The previously developed method using age ratios was employed for the primary analysis. The changes of the EEG activity during the observation period were evaluated statistically. Differences between consecutive EEGs could be observed even in normal subjects but the number of 'false positive' findings was reduced by the automatic data selection, by the choice of proper measures, and by adequate statistical treatment. The diagnostic value of the method was tested in 59 EEGs from 13 patients with well-defined brain disorders. Most EEG changes which could be attributed to the development of the disease were discovered and labelled as statistically significant.