[Time variant bispectral analysis based on adaptive recursive Fourier transformation]. / Zeitvariante Bispektralanalyse auf der Basis einer adaptiv rekursiven Fouriertransformation.

An appropriate investigation of quadratic phase couplings (QPC) in non-stationary signals requires time-variant methods of bispectral analysis. A new approach for time-variant estimation of power spectrum and bispectrum based on an adaptively, recursively estimated Fourier transform (ADFT) is presented in this paper. A reduced calculation effort and the possibility of the calculation of the bispectrum for selected frequency triples are important advantages of this method. Because of the recursive calculation, the ADFT is convenient for analysing ongoing signals. This will be demonstrated for simulated and real biomedical signals.

EEG frequency and phase coupling during human information processing.

Neuronal activity during information processing is represented by oscillations within local or widespread neuronal networks. These oscillations may be recorded by the EEG (electroencephalogram). The oscillatory interaction between neuronal ensembles may be at one single frequency or at different frequencies due to non-linear coupling. The investigation of momentary coherence and phase enables the examination of synchronized oscillatory network activity during fast-changing cognitive processes. On this basis information transfer from occipital areas towards frontal areas could be described during processing of visual presented words. Non-linear phase coupling between oscillations with different frequencies during memory processing was detected by means of cross-bicoherence.

[An optimized filter system for eliminating 50 Hz interference from high resolution ECG]. / Optimiertes Filtersystem für die Entfernung von 50-Hz-Einstreuungen aus dem hochaufgelösten EKG.

A 50/60 Hz notch filter system was designed to eliminate powerline interferences from the high-resolution ECG. This special filter causes only minimal distortions of the power spectra and thus permits us to filter high-resolution ECG's without any appreciable changes in the frequency distribution of the original signal. Since the filter is based on an integer coefficient filter technique, the calculation time is relatively short, and the programming effort comparatively low. Three different automatically selected transfer functions ensure that an optimal filtered signal is obtained despite variations in the value of the interference.

[Methodologic studies in dynamic EMG mapping based on the Hilbert transformation]. / Methodische Untersuchungen zum dynamischen EMG-Mapping auf der Basis der Hilbert-Transformation.

On the basis of Discrete Hilbert Transformation spectral parameters for dynamic EMG analysis can be introduced. Within analysis intervals which are used in the same manner as in spectral analysis, the momentary power of EMG frequency bands can be computed as time series of the same interval duration. A close connection exists between momentary power and the current phenomenology of power spectral analysis because the momentary power of a defined frequency band can be seen as the exact dynamic equivalent of the corresponding mean power value deduced from the power spectrum. A multichannel EMG recording (greater than or equal to 16 channels) makes the representation of the topographical distribution of the spectral parameter by a coloured map possible. Additionally, the momentary power can be used to calculate map sequences. In this way, changes of EMG activity can be quantified by map sequences of an arbitrary time resolution. By calculation of momentary frequency via DHT an artefact detection scheme for ECG interference can be suggested. A completion of this methodology can be carried out by applications of adaptive filtration procedures. Using this concept of EMG processing, a new and common methodical basis of EMG power spectral analysis can be introduced.

Use of discrete Hilbert transformation for automatic spike mapping: a methodological investigation.

On the basis of discrete Hilbert transform (DHT) realised by fast Fourier transform (FFT), a new strategy for automatic spike mapping is introduced. The further computation of the EEG time series after DHT results in the time series of the momentary power and the momentary frequency. Both are used for the solution of the main requirements of automatic spike mapping. The spike-mapping concept introduced meets the requirements of efficient automatic spike detection and also has an insensitivity with regard to EMG interference and transient signal components, a frequent cause of false positive detections. Additionally, there are advantages if the momentary power of the spike is mapped instead of the spike potential. The use of momentary power makes a combination of power spectral mapping and spike-mapping strategies possible.

Using discrete Hilbert transformation for interval-related calculation of the respiration rate in neonates.

From the basis of the fast Fourier transformation (FFT), the discrete Hilbert transformation (DHT) is used to compute the instantaneous respiration rate in neonates. This interval-related computation of respiration rate must be combined with a concept of adaptive filtration of the respiratory movements. This strategy is performed by adaptive recursive estimations of mean values with different adaptation constants. Additionally, a frequency band limitation is carried out on the basis of the peak characteristic of the power spectrum (respiratory movements). By means of the adaptive estimation of the variance of respiratory movements, an amplitude-time window is calculated to choose between epochs with breaths and apnoea.

[Adaptative procedures for pre-processing of evoked potentials]. / Adaptive Verfahren zur Vorverarbeitung evozierter Potentiale.

The investigation of evoked potentials requires suitable consideration of physiological and pathophysiological characteristics of spontaneous and evoked electrical activity of the brain. For this purpose a preprocessing strategy based on adaptive recursive estimation of statistical parameters was developed. In this way, artifact handling, classification, filtering and further preprocessing of spontaneous EEG and evoked potentials can be improved.