Quantitative analysis of discontinuous EEG in premature and full-term newborns during quiet sleep.

To assess the spatio-temporal structure of discontinuous EEG tracing in mature and immature newborns, we analysed mean spectral power in frequency bands between 0.8 and 16.8 Hz in 6 full-term newborns and 7 premature newborns < 32 weeks of conceptional age. The most striking results showed a significantly higher mean spectral power for the first half of bursts than for the second half recorded in > 2.8-14.8 Hz frequency bands. This pattern was more pronounced in premature than in full-term newborns. No clear differences were observed in comparisons between the first and the second half of the interburst periods. In addition, as far as mid and high frequency band spectra were considered, the mean spectral power of burst was, in both groups, higher in the right as compared to the left occipital regions.

Use of adaptive Hilbert transformation for EEG segmentation and calculation of instantaneous respiration rate in neonates.

Broad, as well as narrow, band Hilbert transform filters (HTFs) were used as preprocessing units in the analysis of electroencephalogram (EEG) and respiratory movements in neonates. For these applications, new algorithms for the adaptation of the resonance frequency of a narrow-band-pass filter to the actual signal properties on the basis of an analytic filter design were developed. For the segmentation of the discontinuous EEG, the location of the resonance frequency was imbedded into the learning algorithm of a neural network (NN). In such automatic EEG pattern recognition, the detection of spike activity was taken into consideration. The spike detection scheme introduced uses broad-band HTFs as basis units. Additionally, the algorithm for the continuous control of the resonance frequency was applied to achieve the adaptation of the processing unit that performed the calculation of the instantaneous respiration rate, in this framework, a new on-line method for adaptive frequency estimation that is less sensitive to low signal-to-noise ratios (SNRs) was obtained. The new approaches introduced were tested in comparison with processing methods that have been established for the analysis of experimental and clinical data.

Application of optimized pattern recognition units in EEG analysis: common optimization of preprocessing and weights of neural networks as well as structure optimization.

The main goal of this study is to demonstrate the possibility of training the Neural Network (multilayer perceptron) classifier and preprocessing units simultaneously, i.e., that properties of preprocessing are chosen automatically during the training phase. In the first realization step, adaptive recursive estimation of the power within a frequency band was used as a preprocessing unit. To improve the efficiency of special units, the power and momentary frequency estimation was replaced by methods that are based on adaptive Hilbert transformers. The strategy was developed to obtain optimized recognition units that can be efficiently integrated into strategies for monitoring the cerebral status of neonates. Therefore, applications (e.g., in neonatal EEG pattern recognition) will be shown. Additionally, a method of minimizing the error function was used, where this minimization is based on optimizing the network structure. The results of structure optimization in the field of EEG pattern recognition in epileptic patients can be demonstrated.