Karel Paul

Multivariate Analysis of Full-Term Neonatal Polysomnographic Data

Polysomnography (PSG) is one of the most important noninvasive methods for studying maturation of the child brain. Sleep in infants is significantly different from sleep in adults. This paper addresses the problem of computer analysis of neonatal polygraphic signals. We applied methods designed for differentiating three important neonatal behavioral states: quiet sleep, active sleep, and wakefulness. The proportion of these states is a significant indicator of the maturity of the newborn brain in clinical practice. In this study, we used data provided by the Institute for Care of Mother and Child, Prague (12 newborn infants of similar postconceptional age). The data were scored by an experienced physician to four states (wake, quiet sleep, active sleep, movement artifact). For accurate classification, it was necessary to determine the most informative features. We used a method based on power spectral density (PSD) applied to each EEG channel. We also used features derived from electrooculogram (EOG), electromyogram (EMG), ECG, and respiration [pneumogram (PNG)] signals. The most informative feature was the measure of regularity of respiration from the PNG signal. We designed an algorithm for interpreting these characteristics. This algorithm was based on Markov models. The results of automatic detection of sleep states were compared to the ldquosleep profilesrdquo determined visually. We evaluated both the success rate and the true positive rate of the classification, and statistically significant agreement of the two scorings was found. Two variants, for learning and for testing, were applied, namely learning from the data of all 12 newborns and tenfold cross-validation, and learning from the data of 11 newborns and testing on the data from the 12th newborn. We utilized information obtained from several biological signals (EEG, ECG, PNG, EMG, EOG) for our final classification. We reached the final success rate of 82.5%. The true positive rate was 81.8% and the false positive rate was 6.1%. The most important step in the whole process is feature extraction and feature selection. In this process, we used visualization as an additional tool that helped us to decide which features to select. Proper selection of features may significantly influence the success rate of the classification. We made a visual comparison of the computed features with the manual scoring provided by the expert. A hidden Markov model was used for classification. The advantage of this model is that it determines the future behavior of the process by its present state. In this way, it preserves information about temporal development.

Automatic Detection of Sleep Stages in Neonatal EEG Using the Structural Time Profiles

The new method for automatic sleep stages detection in neonatal EEG was developed. The procedure is based on processing of time profiles computed by adaptive segmentation and subsequent classification of signal graphoelements. The time profiles, functions of the class membership in the course of time, reflect the dynamic EEG structure and may be used for indication of changes in the neonatal sleep stages

Neonatal EEG sleep stages modelling by temporal profiles

The paper deals with the application of the EEG temporal profiles for the neonatal sleep stages modelling. The temporal profiles created by adaptive segmentation and cluster analysis reflect the time structure of the EEG during different periods of sleep. They can be used for neonatal EEG quantification and for the detection of sleep stage changes.

Estimation of the Time Delay of Epileptic Spikes by ICA

The contribution concentrates on application of Independent Component Analysis (ICA) for the detection of small time delays of epileptic spikes in electroencephalographic (EEG) recordings. The ICA method isolates spikes activity by decomposing the input EEG record into independent components. Some of them contain epileptic spikes. ICA detects the time delay of epileptic spikes between channels by separating the epileptic spikes into two or more components. We propose a method of epileptic focus location by ICA from EEG recordings which contain epileptic spikes. The analysis allows presentation of the results in the form of topographic maps. The method was tested on real EEG background signal with artificially simulated epileptic spikes and on EEG records containing real epileptic activity, obtained in four epileptic patients. The tests were used for a comparison with the results of a visual analysis. The tests confirmed a satisfactory agreement between computerized and visual assessments.

52. Non-linear analysis of newborn eeg

Introduction Electroencephalography (EEG) is the measurement of brain electrical activity by means of electrodes positioned on the scalp, which has many important applications in medicine. From one side, visual inspection of EEG signal by neurologists is time consuming, tedious, based on the previous experience and subjective. From the other side, automated classification of EEG signals is very difficult task, as these signals can be noisy and especially when they are recorded during a long time period. In most cases, the agreement of an automatic method with visual analysis is a basis criterion for its acceptance. Today, as well as in the previous decade, a variety of signal processing techniques is being applied on EEG data. Methods In this contribution we present briefly a comprehensive methodology for automatic recognition of behavioral states in neonatal sleep EEG. The methodology is based on segmentation, comprehensive feature extraction and classification of signal segments by supervised learning techniques. The attention was focused on data representation stage in the multistage processing system, namely representation of signal by extracted features. This stage is very important in the analysis of EEG signals in the computational data processing, as it directly affects the classification accuracy. Specifically extracted nonlinear features, whose classification potential was tested, were Hurst exponent and approximate entropy, calculated both for the raw signal and signal after the application of wavelet transform. The methodology was optimized for EEG signal processing in the field of sleep studies in newborns and verified on real clinical neonatal data. Conclusions Based on the obtained results, it can be concluded that the Hurst exponent can be used in the field of neonatal sleep EEG analysis, as it can differentiate well between quiet and active sleep stages. The combination of wavelet transform and approximate entropy were used for the first time in the proposed system for the classification of sleep stages in newborns. The results show that this approach can be used for neonatal sleep EEG analysis, as it provides high classification accuracy. Thus this work provides a reference for enhancing the differentiation of individual neurological states and for the improvement of existing approaches.

Detection of Sleep Stages in Temporal Profiles in Neonatal EEG—k-NN versus k-Means Approach: A Feasibility Study

The aim of this feasibility study is to experimentally verify the detection of changes of sleep stages in neonates with our proposed semi-automated approach using k-NN classification in comparison with a fully automated approach using simple k-means cluster analysis for classification (instead of k-NN). Our semi-automatic approach uses the k-NN classifier trained on etalons (prototypes) created by semi-automated etalons extraction (k-means for etalons suggestion and expert-in-the-loop for verification). Both methods are compared to labelling of sleep stages made by an experienced physician Dr. K. Paul. An EEG recording of full-term neonate is chosen from group of EEG recordings: full-term and preterm neonates recorded from eight electrodes positioned in standard conditions. The EEG recording is digitally preprocessed by mean-removal filter (no other filters are applied) and segmented adaptively. For each segment, 24 features are extracted and send to two classification processes: k-means and k-NN. Classified segments are plotted in temporal profiles (class membership in time) that are analysed for sleep stages using our method of creating a single detection curve from all channels and a threshold is applied on this detection curve to detect sleep stages.

Detection of sleep stages in neonatal EEG records

The aim of this study is the detection of changes in sleep stages in EEG recordings in full-term and preterm newborns. We use a k-NN algorithm as a method of classification. The novelty of our approach lies in semi-automatic etalon (prototype) selection with combination of temporal analysis for sleep stages detection. The semi-automated etalon extraction includes the k-means algorithm for etalons suggestion and an expert-in-the-loop for verification of these etalons. The semi-automated approach improves significantly the time spent on the etalon selection (extraction) by the expert. The whole procedure of EEG signal processing consists of adaptive segmentation, feature extraction, semi-automatic etalon selection using k-means and expert-in-the-loop, classification using k-NN algorithm and temporal profile analysis that is able to detect the neonatal sleep stages for the full-term and even for the preterm neonates, which makes it a unique detection method.

21. Analysis of newborn EEG recorded at 40th week of postconceptional age

19. Default mode network and extrastriate visual resting state network in patients with Parkinson’s disease dementia—I. Rektorova, L. Krajcovicova, R. Marecek, M. Mikl (Applied Neurosciences Research Group, Central European Institute of Technology, CEITEC, Masaryk University, Brno, Czech Republic, First Department of Neurology, Masaryk University and St. Anne’s Hospital, Brno, Czech Republic)