Sleep spindle detection using multivariate Gaussian mixture models.

In this research study we have developed a clustering-based automatic sleep spindle detection method that was evaluated on two different databases. The databases consisted of 20 all-night polysomnograph recordings. Past detection methods have been based on subject-independent and some subject-dependent parameters, such as fixed or variable thresholds to identify spindles. Using a multivariate Gaussian mixture model clustering technique, our algorithm was developed to use only subject-specific parameters to detect spindles. We have obtained an overall sensitivity range (65.1-74.1%) at a (59.55-119.7%) false positive proportion.

Blood pressure and cardiovascular parameters during sleep arousals.

Sleep arousal is generally known as a transient episode of wakefulness into the sleepiness. Sleep arousals can be classified based on their association and accompany with pathological episodes. In this paper, our objective was to find out whether various types of sleep arousals influence on blood pressure and Heart Rate Variability (HRV). We analysed continuous Diastolic and Systolic Blood Pressures (DBP and SBP), Pulse Transit Time (PTT) as well as High and Low Frequency components (HF and LF) of HRV in different types of arousals. We developed Slope Index (SI) to determine whether a feature was ascending or descending before, during and after the occurrence of a sleep arousal. Slope Index Positive Percentage (SIPP) was created and computed for all features to find out the percentage of arousals with an ascending trend of a cardiovascular feature. In pre-arousal epochs, we obtained SIPPDBP= 48.9%, SIPPSBP = 48.2% and SIPPLF = 41%. Whilst during the arousal episodes, the SIPPDBP, SIPPSBP and SIPPLF increased to 57.2%, 57.4% and 78.9%, correspondingly. This means that during arousal occurrence these parameters were likelier to rise. Whereas SIPP of PTT and HF component of HRV during arousals were less than prearousal. This indicated PTT and HF were highly probable to drop during the arousal than to rise. The high SIPPDBP and SIPPSBP parameters, approximately 76%, during the arousals indicates that sleep arousals may cause a sudden increase in blood pressure.

Insomnia Characterization: From Hypnogram to Graph Spectral Theory.

OBJECTIVE: To quantify and differentiate control and insomnia sleep onset patterns through biomedical signal processing of overnight polysomnograms. METHODS: The approach consisted of three tandem modules: 1) biosignal processing module, which used state-space time-varying autoregressive moving average (TVARMA) processes with recursive particle filter, 2) hypnogram generation module that implemented a fuzzy inference system (FIS), and 3) insomnia characterization module that discriminated between control and subjects with insomnia using a logistic regression model trained with a set of similarity measures ( d1, d2 , d3, d4). The study employed sleep onset periods from 16 control and 16 subjects with insomnia. RESULTS: State-spaced TVARMA processes with recursive particle filtering provided resilience to nonlinear, nonstationary, and non-Gaussian conditions of biosignals. FIS managed automated sleep scoring robust to intersubjects' and interraters' variability. The similarity distances quantified in a scalar measure the transitions amongst sleep onset stages, computed from expert and automated hypnograms. A statistical set of unpaired two-tailed t -tests suggested that distances d1 , d2, and d3 had larger statistical significance ( ) to characterize sleeping patterns. The logistic regression model classified control and subjects with insomnia with sensitivity 87 % , specificity 75 %, and accuracy 81 %. CONCLUSION: Our approach can perform a supportive role in either biosignal processing, sleep staging, insomnia characterization, or all the previous, coping with time-consuming procedures and massive data volumes of standard protocols. SIGNIFICANCE: The introduction of graph spectral theory and logistic regression for the diagnosis of insomnia represents a novel concept, attempting to describe complex sleep dynamics throughout transitions networks and scalar measures.

Sleep electroencephalography and heart rate variability interdependence amongst healthy subjects and insomnia/schizophrenia patients.

The quantification of interdependencies within autonomic nervous system has gained increasing importance to characterise healthy and psychiatric disordered subjects. The present work introduces a biosignal processing approach, suggesting a computational resource to estimate coherent or synchronised interactions as an eventual supportive aid in the diagnosis of primary insomnia and schizophrenia pathologies. By deploying linear, nonlinear and statistical methods upon 25 electroencephalographic and electrocardiographic overnight sleep recordings, the assessment of cross-correlation, wavelet coherence and [Formula: see text]:[Formula: see text] phase synchronisation is focused on tracking discerning features amongst the clinical cohorts. Our results indicate that certain neuronal oscillations interact with cardiac power bands in distinctive ways responding to standardised sleep stages and patient groups, which promotes the hypothesis of subtle functional dynamics between neuronal assembles and (para)sympathetic activity subject to pathophysiological conditions.

Photoplethysmography derivatives and pulse transit time in overnight blood pressure monitoring.

Overnight continuous blood pressure measurement provides simultaneous monitoring of blood pressure and sleep architecture. By this means, we are able to investigate whether different sleep events are associated to blood pressure fluctuations. In this paper, we used the Pulse Transit Time (PTT) to develop and evaluate functions for measurement of blood pressure. We focused on the first and second derivatives of fingertip Photoplethysmography (PPG) recordings to detect PPG critical points. By applying R wave of ECG and PPG critical points, we created two PTT-based models for estimation of systolic and diastolic blood pressure (SBP and DBP). Seven subjects polysomnography datasets that contained PPG, ECG and blood pressure recordings were utilised to validate and compare developed PTT-BP functions. Results found that if the peak of the first derivative of PPG (VPG) was considered as the pulse pressure arrival point, the resulted PTT (PTTV) would more accurately predict both SBP and DBP. The average R-squared coefficient for SBP and DBP were correspondingly 0.593 and 0.416. The obtained mean error for PTTV based functions in SBP was ±3.96 mmHg with standard deviation of 1.41 mmHg and in DBP was ±6.88 mmHg with standard deviation of 3.03 mmHg. We concluded PTT detected from VPG is a reliable and suitable maker for overnight continuous blood pressure monitoring.

Pulse transit time and heart rate variability in sleep staging.

This paper presents a new and robust algorithm for detection of sleep stages by using the lead I of the Electrocardiography (ECG) and a fingertip Photoplethysmography (PPG) sensor, validated using multiple overnight PSG recordings consisting of 20 human subjects (9 insomniac and 11 healthy). Heart Rate Variability (HRV) and Pulse Transit Time (PTT) biomarkers which were extracted from ECG and PPG biosignals then employed to extract features. Distance Weighted k-Nearest Neighbours (DWk-NN) was used as classifier to differentiate sleep epochs. The validation of the algorithm was evaluated by Leave-One-Out-Cross-Validation method. The average accuracy of 73.4% with standard deviation of 6.4 was achieved while the algorithm could distinguish stages 2, 3 of non-rapid eye movement sleep by average sensitivity of almost 80%. The lowest mean sensitivity of 53% was for stage 1. These results demonstrate that an algorithm based on PTT and HRV spectral analysis is able to classify and distinguish sleep stages with high accuracy and sensitivity. In addition the proposed algorithm is capable to be improved and implemented as a wearable, comfortable and cheap instrument for sleep screening.

Comparison between human awake, meditation and drowsiness EEG activities based on directed transfer function and MVDR coherence methods.

This study examined the electroencephalogram functional connectivity (coherence) and effective connectivity (flow of information) of selected brain regions during three different attentive states: awake, meditation and drowsiness. For the estimation of functional connectivity (coherence), Welch and minimum variance distortionless response (MVDR) methods were compared. The MVDR coherence was found to be more suitable since it is both data and frequency dependent and enables higher spectral resolution, while Welch's periodogram-based approach is both data and frequency independent. The directed transfer function (DTF) method was applied in order to estimate the effective connectivity or brain's flow of information between different regions during each state. DTF enables to identify the main brain areas that initiate EEG activity and the spatial distribution of these activities with time. Analysis was conducted using the EEG data of 30 subjects (ten awake, ten drowsy and ten meditating) focusing on six main electrodes (F3, F4, C3, C4, P3, P4, O1 and O2). For each subject, EEG data were recorded during 5-min baseline and 15 min of a specific condition (awake, meditation or drowsiness). Statistical analysis included the Kruskal-Wallis (KW) nonparametric analysis of variance followed by post hoc tests with Bonferroni alpha correction. The results reveal that both states of drowsiness and meditation states lead to a marked difference in the brain's flow of information (effective connectivity) as shown by DTF analyses. In specific, a significant increase in the flow of information in the delta frequency band was found only in the meditation condition and was further found to originate from frontal (F3, F4), parietal (P3, P4) and occipital (O1, O2) regions. Altogether, these results suggest that a change in attentiveness leads to significant changes in the spectral profile of the brain's information flow as well as in its functional connectivity and that these changes can be captured using coherence and DTF analyses.

Searching arousals: A fuzzy logic approach.

This paper presents a computational approach to detect spontaneous, chin tension and limb movement-related arousals by estimating neuronal and muscular activity. Features extraction is carried out by Time Varying Autoregressive Moving Average (TVARMA) models and recursive particle filtering. Classification is performed by a fuzzy inference system with rule-based decision scheme based upon the AASM scoring rules. Our approach yielded two metrics: arousal density and arousal index to comply with standardised clinical benchmarking. The obtained statistics achieved error deviation around ±1.5 to ±30. These results showed that our system can differentiate amongst 3 different types of arousals, subject to inter-subject variability and up-to-date scoring references.

Characterising insomnia: A graph spectral theory approach.

This paper introduces a computational approach to characterise healthy controls and insomniacs based on graph spectral theory. Based upon expert-generated hypnograms of sleep onset periods, a network of sleep stages transitions is derived to compute four similarity distances amongst subjects' sleeping patterns. A subsequent statistical analysis is performed to differentiate the 16-subject healthy group from a 16-patient disordered cohort. Our findings demonstrated that the similarity distances based on eigenvalues determination, i.e. d1 and d4 were the most reliable and robust measures to characterise insomniacs, discriminating 93% and 87% of the affected population, respectively.

Automated sleep spindle detection using IIR filters and a Gaussian Mixture Model.

Sleep spindle detection using modern signal processing techniques such as the Short-Time Fourier Transform and Wavelet Analysis are common research methods. These methods are computationally intensive, especially when analysing data from overnight sleep recordings. The authors of this paper propose an alternative using pre-designed IIR filters and a multivariate Gaussian Mixture Model. Features extracted with IIR filters are clustered using a Gaussian Mixture Model without the use of any subject independent thresholds. The Algorithm was tested on a database consisting of overnight sleep PSG of 5 subjects and an online public spindles database consisting of six 30 minute sleep excerpts. An overall sensitivity of 57% and a specificity of 98.24% was achieved in the overnight database group and a sensitivity of 65.19% at a 16.9% False Positive proportion for the 6 sleep excerpts.

Linear and non-linear interdependence of EEG and HRV frequency bands in human sleep.

The characterisation of functional interdependencies of the autonomic nervous system (ANS) stands an evergrowing interest to unveil electroencephalographic (EEG) and Heart Rate Variability (HRV) interactions. This paper presents a biosignal processing approach as a supportive computational resource in the estimation of sleep dynamics. The application of linear, non-linear methods and statistical tests upon 10 overnight polysomnographic (PSG) recordings, allowed the computation of wavelet coherence and phase locking values, in order to identify discerning features amongst the clinical healthy subjects. Our findings showed that neuronal oscillations θ, α and σ interact with cardiac power bands at mid-to-high rank of coherence and phase locking, particularly during NREM sleep stages.

Automated Sleep Spindle detection using novel EEG features and mixture models.

Research in automated Sleep Spindle detection has been highly explored in the past few years. Although a number of automated techniques were developed, many of them were based on using fixed parameters or thresholds which do not consider subject specific differences. In this research study, we introduce a novel method of sleep spindle detection using Gaussian Mixture Models with no fixed parameters or thresholds. The algorithm was tested on an online public spindles database consisting of six 30 minute sleep excerpts extracted from whole night recordings of 6 subjects. The results obtained were better when compared with other methods. We obtained an overall sensitivity of 74.9% at a 28% False Positive proportion.

Information flow and coherence of EEG during awake, meditation and drowsiness.

A comparison of coupling (information flow) and coherence (connectedness) of the brain regions between human awake, meditation and drowsiness states was carried out in this study. The Directed Transfer Function (DTF) method was used to estimate the coupling or brain's flow of information between different regions during each condition. Welch and Minimum Variance Distortionless Response (MVDR) methods were utilised to estimate the coherence between brain areas. Analysis was conducted using the EEG data of 30 subjects (10 awake, 10 drowsiness and 10 meditating) with 6 EEG electrodes. The EEG data was recorded for each subject during 5 minutes baseline and 15 minutes of three specific conditions (awake, meditation or drowsiness). Statistical analysis was carried out which consisted of the Kruskal-Wallis (KW) non-parametric analysis of variance followed by post-hoc tests with Bonferroni alpha-correction. The results of this study revealed that a change in external awareness led to substantial differences in the spectral profile of the brain's information flow as well as it's connectedness.

A single-trial toolbox for advanced sleep polysomnographic preprocessing.

The application of polysomnographic (PSG) studies for monitoring sleep activity is a multi-parametric practice that involves a diverse group of biological signals. A suitable preprocessing of such signals assures a more profitable feature extraction and classification operations. Therefore, the proposed preprocessing toolbox performs segmentation, filtering, denoising, whitening and artefact removal tasks upon multi-channel PSG recordings. In order to assess toolbox's efficiency, clinical experiments are conducted, as well as, quantitative and qualitative metrics are discussed. Our findings reveal outperforming efficiency by artefacts and noise rejection after single-trial and multi-stage preprocessing.

Automatic detection of the wake and stage 1 sleep stages using the EEG sub-epoch approach.

Studies by Rechtschaffen and Kales (R&K), rely on 30-sec epochs to score sleep stages. In this paper, we introduce a new approach based on three consecutive and non-consecutive 6-sec sub-epochs for the detection of the wake stage and stage 1 sleep. The Relative Spectral Energy Band (RSEB) is used as a feature extraction from the electroencephalographic (EEG) signal. Spectral estimation is performed using non-parametric and parametric methods. We then compared the performance of the conventional 30-sec epochs with the three consecutive and non-consecutive 6-sec epochs. The outcomes of this study showed that while the accuracy varies between subjects, the non-parametric method proved to be more effective with stage 1 sleep detection and the parametric method was more effective for wake stage detection. The non-consecutive sub-epoch method was more effective and consecutive method was least effective in non-parametric stage 1 detection. Alternatively, the 30-second epoch method was most effective for parametric wake stage detection.

Evaluation of FCS self and peer-assessment approach based on Cooperative and Engineering Design learning.

The Cooperative Learning in Engineering Design curriculum can be enhanced with structured and timely self and peer assessment teaching methodologies which can easily be applied to any Biomedical Engineering curriculum. A study was designed and implemented to evaluate the effectiveness of this structured and timely self and peer assessment on student team-based projects. In comparing the 'peer-blind' and 'face-to-face' Fair Contribution Scoring (FCS) methods, both had advantages and disadvantages. The 'peer-blind' self and peer assessment method would cause high discrepancy between self and team ratings. But the 'face-to-face' method on the other hand did not have the discrepancy issue and had actually proved to be a more accurate and effective, indicating team cohesiveness and good cooperative learning.

The dark side of the alpha rhythm: fMRI evidence for induced alpha modulation during complete darkness.

The unique role of the EEG alpha rhythm in different states of cortical activity is still debated. The main theories regarding alpha function posit either sensory processing or attention allocation as the main processes governing its modulation. Closing and opening eyes, a well-known manipulation of the alpha rhythm, could be regarded as attention allocation from inward to outward focus though during light is also accompanied by visual change. To disentangle the effects of attention allocation and sensory visual input on alpha modulation, 14 healthy subjects were asked to open and close their eyes during conditions of light and of complete darkness while simultaneous recordings of EEG and fMRI were acquired. Thus, during complete darkness the eyes-open condition is not related to visual input but only to attention allocation, allowing direct examination of its role in alpha modulation. A data-driven ridge regression classifier was applied to the EEG data in order to ascertain the contribution of the alpha rhythm to eyes-open/eyes-closed inference in both lighting conditions. Classifier results revealed significant alpha contribution during both light and dark conditions, suggesting that alpha rhythm modulation is closely linked to the change in the direction of attention regardless of the presence of visual sensory input. Furthermore, fMRI activation maps derived from an alpha modulation time-course during the complete darkness condition exhibited a right frontal cortical network associated with attention allocation. These findings support the importance of top-down processes such as attention allocation to alpha rhythm modulation, possibly as a prerequisite to its known bottom-up processing of sensory input.

Cross-correlation of EEG frequency bands and heart rate variability for sleep apnoea classification.

Sleep apnoea is a sleep breathing disorder which causes changes in cardiac and neuronal activity and discontinuities in sleep pattern when observed via electrocardiogram (ECG) and electroencephalogram (EEG). Using both statistical analysis and Gaussian discriminative modelling approaches, this paper presents a pilot study of assessing the cross-correlation between EEG frequency bands and heart rate variability (HRV) in normal and sleep apnoea clinical patients. For the study we used EEG (delta, theta, alpha, sigma and beta) and HRV (LF(nu), HF(nu) and LF/HF) features from the spectral analysis. The statistical analysis in different sleep stages highlighted that in sleep apnoea patients, the EEG delta, sigma and beta bands exhibited a strong correlation with HRV features. Then the correlation between EEG frequency bands and HRV features were examined for sleep apnoea classification using univariate and multivariate Gaussian models (UGs and MGs). The MG outperformed the UG in the classification. When EEG and HRV features were combined and modelled with MG, we achieved 64% correct classification accuracy, which is 2 or 8% improvement with respect to using only EEG or ECG features. When delta and acceleration coefficients of the EEG features were incorporated, then the overall accuracy improved to 71%.

Alterations in sleep EEG activity during the hypopnoea episodes.

The Obstructive Sleep Apnoea Hypopnoea Syndrome (OSAH) means "cessation of breath" during the sleep hours and the sufferers often experience related changes in the electrical activity of the brain and heart. The aim of this paper is to investigate any possible changes in the human electroencephalographic (EEG) activity due to hypopnoea (mild case of cessation of breath) occurrences by applying the non-linear and linear time series methods. The results from this study indicated significant changes in the human EEG activity due to hypopnoea episodes by applying the non-linear, Lyapunov exponent method at C3 EEG electrode site. This non-linear method can be applied in future evaluation of sleep EEG transients during the OSAH episodes.

Correlation of sleep EEG frequency bands and Heart Rate Variability.

Sleep apnoea is a sleep breathing disorder which causes changes in cardiac and neuronal activity and discontinuities in sleep pattern when observed via electrocardiogram (ECG) and electroencephalogram (EEG). This paper presents a pilot study result of assessing the correlation between EEG frequency bands and ECG Heart Rate Variability (HRV) in normal and sleep apnoea human clinical patients at different sleep stages. In sleep apnoea patients, the results have shown that EEG delta, sigma and beta bands exhibited a strong correlation with cardiac HRV parameters at different sleep stages.