Using the Hilbert-Huang transform to measure the electroencephalographic effect of propofol

SQI-DOANet: electroencephalogram-based deep neural network for estimating signal quality index and depth of anaesthesia

Objective. Monitoring the depth of anaesthesia (DOA) during surgery is of critical importance. However, during surgery electroencephalography (EEG) is usually subject to various disturbances that affect the accuracy of DOA. Therefore, accurately estimating noise in EEG and reliably assessing DOA remains an important challenge. In this paper, we proposed a signal quality index (SQI) network (SQINet) for assessing the EEG signal quality and a DOA network (DOANet) for analyzing EEG signals to precisely estimate DOA. The two networks are termed SQI-DOANet.Approach. The SQINet contained a shallow convolutional neural network to quickly determine the quality of the EEG signal. The DOANet comprised a feature extraction module for extracting features, a dual attention module for fusing multi-channel and multi-scale information, and a gated multilayer perceptron module for extracting temporal information. The performance of the SQI-DOANet model was validated by training and testing the model on the large VitalDB database, with the bispectral index (BIS) as the reference standard.Main results. The proposed DOANet yielded a Pearson correlation coefficient with the BIS score of 0.88 in the five-fold cross-validation, with a mean absolute error (MAE) of 4.81. The mean Pearson correlation coefficient of SQI-DOANet with the BIS score in the five-fold cross-validation was 0.82, with an MAE of 5.66.Significance. The SQI-DOANet model outperformed three compared methods. The proposed SQI-DOANet may be used as a new deep learning method for DOA estimation. The code of the SQI-DOANet will be made available publicly athttps://github.com/YuRui8879/SQI-DOANet.

Changes in EEG frequency characteristics during sevoflurane general anesthesia: feature extraction by variational mode decomposition

Mode decomposition is a method for extracting the characteristic intrinsic mode function (IMF) from various multidimensional time-series signals. Variational mode decomposition (VMD) searches for IMFs by optimizing the bandwidth to a narrow band with the [Formula: see text] norm while preserving the online estimated central frequency. In this study, we applied VMD to the analysis of electroencephalogram (EEG) recorded during general anesthesia. Using a bispectral index monitor, EEGs were recorded from 10 adult surgical patients (the median age: 47.0, and the percentile range: 27.0-59.3 years) who were anesthetized with sevoflurane. We created an application named EEG Mode Decompositor, which decomposes the recorded EEG into IMFs and displays the Hilbert spectrogram. Over the 30-min recovery from general anesthesia, the median (25-75 percentile range) bispectral index increased from 47.1 (42.2-50.4) to 97.4 (96.5-97.6), and the central frequencies of IMF-1 showed a significant change from 0.4 (0.2-0.5) Hz to 0.2 (0.1-0.3) Hz. IMF-2, IMF-3, IMF-4, IMF-5, and IMF-6 increased significantly from 1.4 (1.2-1.6) Hz to 7.5 (1.5-9.3) Hz, 6.7 (4.1-7.6) Hz to 19.4 (6.9-20.0) Hz, 10.9 (8.8-11.4) Hz to 26.4 (24.2-27.2) Hz, 13.4 (11.3-16.6) Hz to 35.6 (34.9-36.1) Hz, and 12.4 (9.7-18.1) Hz to 43.2 (42.9-43.4) Hz, respectively. The characteristic frequency component changes in specific IMFs during emergence from general anesthesia were visually captured by IMFs derived using VMD. EEG analysis by VMD is useful for extracting distinct changes during general anesthesia.

Time-trend analysis of the center frequency of the intrinsic mode function from the Hilbert-Huang transform of electroencephalography during general anesthesia: a retrospective observational study

BACKGROUND

Anesthesiologists are required to maintain an optimal depth of anesthesia during general anesthesia, and several electroencephalogram (EEG) processing methods have been developed and approved for clinical use to evaluate anesthesia depth. Recently, the Hilbert-Huang transform (HHT) was introduced to analyze nonlinear and nonstationary data. In this study, we assessed whether the changes in EEG characteristics during general anesthesia that are analyzed by the HHT are useful for monitoring the depth of anesthesia.

METHODS

This retrospective observational study enrolled patients who underwent propofol anesthesia. Raw EEG signals were obtained from a monitor through a previously developed software application. We developed an HHT analyzer to decompose the EEG signal into six intrinsic mode functions (IMFs) and estimated the instantaneous frequencies (HHT_IF) for each IMF. Changes over time in the raw EEG waves and parameters such as HHT_IF, BIS, spectral edge frequency 95 (SEF95), and electromyogram parameter (EMGlow) were assessed, and a Gaussian process regression model was created to assess the association between BIS and HHT_IF.

RESULTS

We analyzed EEG signals from 30 patients. The beta oscillation frequency range (13-25 Hz) was detected in IMF1 and IMF2 during the awake state, then after loss of consciousness, the frequency decreased and alpha oscillation (8-12 Hz) was detected in IMF2. At the emergence phase, the frequency increased and beta oscillations were detected in IMF1, IMF2, and IMF3. BIS and EMGlow changed significantly during the induction and emergence phases, whereas SEF95 showed a wide variability and no significant changes during the induction phase. The root mean square error between the observed BIS values and the values predicted by a Gaussian process regression model ranged from 4.69 to 9.68.

CONCLUSIONS

We applied the HHT to EEG analyses during propofol anesthesia. The instantaneous frequency in IMF1 and IMF2 identified changes in EEG characteristics during induction and emergence from general anesthesia. Moreover, the HHT_IF in IMF2 showed strong associations with BIS and was suitable for depicting the alpha oscillation. Our study suggests that the HHT is useful for monitoring the depth of anesthesia.

The influence of moderate or deep neuromuscular block status on anesthetic depth monitoring system during total intravenous anesthesia using propofol and remifentanil: A randomized trial

The neuromuscular block state may affect the electroencephalogram-derived index representing the anesthetic depth. We applied an Anesthetic Depth Monitoring for Sedation (ADMS) to patients undergoing laparoscopic cholecystectomy under total intravenous anesthesia, and evaluated the requirement of propofol according to the different neuromuscular block state. Adult patients scheduled to undergo laparoscopic cholecystectomy were enrolled and randomly assigned to either the moderate (MB) or deep neuromuscular block (DB) group. The UniCon sensor of ADMS was applied to monitor anesthetic depth and the unicon value was maintained between 40 and 50 during the operation. According to the group assignment, intraoperative rocuronium was administered to maintain proper neuromuscular block state, moderate or deep block state. The unicon value, electromyography (EMG) index, and total dose of propofol and rocuronium were analyzed. At similar anesthetic depth, less propofol was used in the DB group compared to the MB group (6.19 ± 1.36 in the MB mg/kg/h group vs 4.93 ± 3.02 mg/kg/h in the DM group, p = 0.042). As expected, more rocuronium were used in the DB group than in the MB group (0.8 ± 0.2 mg/kg in the MB group vs 1.2 ± 0.2 mg/kg in the DB group, p = 0.023) and the EMG indices were lower in the DB group than in the MB group, at several time points as follows: at starting operation (p < 0.001); at 15 (p = 0.019), 45 (p = 0.011), and 60 min (p < 0.001) after the initiation of the operation; at the end of operation (p = 0.003); and at 5 min after the administration of sugammadex (p < 0.001). At similar anesthetic depth, patients under the deep neuromuscular block state required less propofol with lower intraoperative EMG indices compared to those under the moderate neuromuscular block state during general anesthesia.

Monitoring the level of hypnosis using a hierarchical SVM system

Monitoring level of hypnosis is a major ongoing challenge for anesthetists to reduce anesthetic drug consumption, avoiding intraoperative awareness and prolonged recovery. This paper proposes a novel automated method for accurate assessing of the level of hypnosis with sevoflurane in 17 patients using the electroencephalogram signal. In this method, a set of distinctive features and a hierarchical classification structure based on support vector machine (SVM) methods, is proposed to discriminate the four levels of anesthesia (awake, light, general and deep states). The first stage of the hierarchical SVM structure identifies the awake state by extracting Shannon Permutation Entropy, Detrended Fluctuation Analysis and frequency features. Then deep state is identified by extracting the sample entropy feature; and finally light and general states are identified by extracting the three mentioned features of the first step. The accuracy of the proposed method of analyzing the brain activity during anesthesia is 94.11%; which was better than previous studies and also a commercial monitoring system (Response Entropy Index).

An EEG-Based Attentiveness Recognition System Using Hilbert–Huang Transform and Support Vector Machine

Purpose

Attentiveness recognition benefits the detection of the mental state and concentration when humans perform specific tasks. Hilbert–Huang transform (HHT) is useful for the analysis of nonlinear or nonstationary bio-signals including brainwaves. In this work, a method is proposed for the characterization of attentiveness levels by using electroencephalogram (EEG) signals and HHT analysis.

Methods

Single-channel EEG signals from the frontal area were acquired from participants at different levels of attentiveness and were decomposed into a set of intrinsic mode functions (IMF) by empirical mode decomposition (EMD). Hilbert transform analysis was applied to each IMF to obtain the marginal frequency spectrum. Then the band powers and spectral entropies (SEs) were selected as the attributes of a support vector machine (SVM) for a two-class classification task.

Results

Compared with the predictive models of approximate entropy (ApEn) and fast Fourier transform (FFT), the results show that the band powers extracted from IMF2 to IMF5 of $$\alpha$$α and $$\beta$$β waves and their SE can best discriminate between attentive and relaxed states with the average classification accuracy of 84.80%.

Conclusion

In conclusion, this integrated signal processing method is capable of attentiveness recognition that can offer efficient differentiation and may be used in a clinical setting for the detection of attention deficit.

Frontal-temporal functional connectivity of EEG signal by standardized permutation mutual information during anesthesia

Quantifying brain dynamics during anesthesia is an important challenge for understanding the neurophysiological mechanisms of anesthetic drug effect. Several single channel Electroencephalogram (EEG) indices have been proposed for monitoring anesthetic drug effect. The most commonly used single channel commercial index is the Bispectral index (BIS). However, this monitor has shown some drawbacks. In this study, a nonlinear functional connectivity measure named Standardized Permutation Mutual Information (SPMI) is proposed to describe communication between two-channel EEG signals at frontal and temporal brain regions during a controlled propofol-induced anesthesia and recovery design from eight subjects. The SPMI index has higher correlation with estimated propofol effect-site concentration and has better ability to distinguish three anesthetic states of patient than the other functional connectivity indexes (cross-correlation, coherence, phase analysis) and also the BIS index. Moreover, the SPMI index has a faster reaction to the effect of drug concentration, less variability at the consciousness state and better robustness to noise than BIS.

Use of Multiple EEG Features and Artificial Neural Network to Monitor the Depth of Anesthesia

The electroencephalogram (EEG) can reflect brain activity and contains abundant information of different anesthetic states of the brain. It has been widely used for monitoring depth of anesthesia (DoA). In this study, we propose a method that combines multiple EEG-based features with artificial neural network (ANN) to assess the DoA. Multiple EEG-based features can express the states of the brain more comprehensively during anesthesia. First, four parameters including permutation entropy, 95% spectral edge frequency, BetaRatio and SynchFastSlow were extracted from the EEG signal. Then, the four parameters were set as the inputs to an ANN which used bispectral index (BIS) as the reference output. 16 patient datasets during propofol anesthesia were used to evaluate this method. The results indicated that the accuracies of detecting each state were 86.4% (awake), 73.6% (light anesthesia), 84.4% (general anesthesia), and 14% (deep anesthesia). The correlation coefficient between BIS and the index of this method was 0.892 (p<0.001). The results showed that the proposed method could well distinguish between awake and other anesthesia states. This method is promising and feasible for a monitoring system to assess the DoA.

Non-linear Entropy Analysis in EEG to Predict Treatment Response to Repetitive Transcranial Magnetic Stimulation in Depression

Background: Biomarkers that predict clinical outcomes in depression are essential for increasing the precision of treatments and clinical outcomes. The electroencephalogram (EEG) is a non-invasive neurophysiological test that has promise as a biomarker sensitive to treatment effects. The aim of our study was to investigate a novel non-linear index of resting state EEG activity as a predictor of clinical outcome, and compare its predictive capacity to traditional frequency-based indices.

Methods: EEG was recorded from 62 patients with treatment resistant depression (TRD) and 25 healthy comparison (HC) subjects. TRD patients were treated with excitatory repetitive transcranial magnetic stimulation (rTMS) to the dorsolateral prefrontal cortex (DLPFC) for 4 to 6 weeks. EEG signals were first decomposed using the empirical mode decomposition (EMD) method into band-limited intrinsic mode functions (IMFs). Subsequently, Permutation Entropy (PE) was computed from the obtained second IMF to yield an index named PEIMF2. Receiver Operator Characteristic (ROC) curve analysis and ANOVA test were used to evaluate the efficiency of this index (PEIMF2) and were compared to frequency-band based methods.

Results: Responders (RP) to rTMS exhibited an increase in the PEIMF2 index compared to non-responders (NR) at F3, FCz and FC3 sites (p < 0.01). The area under the curve (AUC) for ROC analysis was 0.8 for PEIMF2 index for the FC3 electrode. The PEIMF2 index was superior to ordinary frequency band measures.

Conclusion: Our data show that the PEIMF2 index, yields superior outcome prediction performance compared to traditional frequency band indices. Our findings warrant further investigation of EEG-based biomarkers in depression; specifically entropy indices applied in band-limited EEG components. Registration in ClinicalTrials.Gov; identifiers NCT02800226 and NCT01887782.

Monitoring the Depth of Anesthesia Using a New Adaptive Neurofuzzy System

Accurate and noninvasive monitoring of the depth of anesthesia (DoA) is highly desirable. Since the anesthetic drugs act mainly on the central nervous system, the analysis of brain activity using electroencephalogram (EEG) is very useful. This paper proposes a novel automated method for assessing the DoA using EEG. First, 11 features including spectral, fractal, and entropy are extracted from EEG signal and then, by applying an algorithm according to exhaustive search of all subsets of features, a combination of the best features (Beta-index, sample entropy, shannon permutation entropy, and detrended fluctuation analysis) is selected. Accordingly, we feed these extracted features to a new neurofuzzy classification algorithm, adaptive neurofuzzy inference system with linguistic hedges (ANFIS-LH). This structure can successfully model systems with nonlinear relationships between input and output, and also classify overlapped classes accurately. ANFIS-LH, which is based on modified classical fuzzy rules, reduces the effects of the insignificant features in input space, which causes overlapping and modifies the output layer structure. The presented method classifies EEG data into awake, light, general, and deep states during anesthesia with sevoflurane in 17 patients. Its accuracy is 92% compared to a commercial monitoring system (response entropy index) successfully. Moreover, this method reaches the classification accuracy of 93% to categorize EEG signal to awake and general anesthesia states by another database of propofol and volatile anesthesia in 50 patients. To sum up, this method is potentially applicable to a new real-time monitoring system to help the anesthesiologist with continuous assessment of DoA quickly and accurately.

Reprint of "A new approach to analyze data from EEG-based concealed face recognition system"

The purpose of this study is to extend a feature set with non-linear features to improve classification rate of guilty and innocent subjects. Non-linear features can provide extra information about phase space. The Event-Related Potential (ERP) signals were recorded from 49 subjects who participated in concealed face recognition test. For feature extraction, at first, several morphological characteristics, frequency bands, and wavelet coefficients (we call them basic-features) are extracted from each single-trial ERP. Recurrence Quantification Analysis (RQA) measures are then computed as non-linear features from each single-trial. We apply Genetic Algorithm (GA) to select the best feature set and this feature set is used for classification of data using Linear Discriminant Analysis (LDA) classifier. Next, we use a new approach to improve classification results based on introducing an adaptive-threshold. Results indicate that our method is able to correctly detect 91.83% of subjects (45 correct detection of 49 subjects) using combination of basic and non-linear features, that is higher than 87.75% for basic and 79.59% for non-linear features. This shows that combination of non-linear and basic- features could improve classification rate.

A new approach to analyze data from EEG-based concealed face recognition system

The purpose of this study is to extend a feature set with non-linear features to improve classification rate of guilty and innocent subjects. Non-linear features can provide extra information about phase space. The Event-Related Potential (ERP) signals were recorded from 49 subjects who participated in concealed face recognition test. For feature extraction, at first, several morphological characteristics, frequency bands, and wavelet coefficients (we call them basic-features) are extracted from each single-trial ERP. Recurrence Quantification Analysis (RQA) measures are then computed as non-linear features from each single-trial. We apply Genetic Algorithm (GA) to select the best feature set and this feature set is used for classification of data using Linear Discriminant Analysis (LDA) classifier. Next, we use a new approach to improve classification results based on introducing an adaptive-threshold. Results indicate that our method is able to correctly detect 91.83% of subjects (45 correct detection of 49 subjects) using combination of basic and non-linear features, that is higher than 87.75% for basic and 79.59% for non-linear features. This shows that combination of non-linear and basic- features could improve classification rate.

Comparison of FFT and marginal spectra of EEG using empirical mode decomposition to monitor anesthesia

BACKGROUND AND OBJECTIVE

Intraoperative awareness refers that patients can recall aspects of their surgery after being put under general anesthesia. This distressing complication causes affected patients to be conscious and probably feel pain, leading to emotional trauma or other sequelae. Monitoring and administrating the depth of anesthesia is necessary to prevent patients from awareness during a medical operation. In this paper, we analyzed the electroencephalograms (EEGs) of patients to characterize their anesthesia. The data set, "awareness" and "anesthesia" groups, each contained 558 samples, including patients who had undergone different types of surgeries.

METHODS

EEG signals acquired from patients in an aware state or under anesthesia were decomposed into a set of intrinsic mode functions (IMFs) through empirical mode decomposition (EMD). Fast Fourier transform (FFT) and Hilbert transform (HT) analyses were then performed on each IMF to determine the frequency spectra. The probability distributions of expected values of frequencies were generated for the same IMF in the two groups of patients. The corresponding statistical data, including analysis of variance tests, were also calculated. A receiver operating characteristic curve was used to identify optimal frequency value to discriminate between the two states of consciousness.

RESULTS

The frequencies of the IMFs for aware patients were found to be higher than those for anesthetized patients. The optimal frequency threshold by using FFT (or HT) for IMF 1 was 21.08 (or 25.00) Hz. IMF1 performed the highest with respect to the area under the curve (AUC) of 0.993 for FFT (or 0.989 for HT); hence it can be applied as a useful classifier to distinguish between fully anesthetized patients and aware patients.

CONCLUSIONS

This paper proposes a method for identifying whether patients' state of consciousness during a range of surgery types is "under anesthesia" or "aware." Our method involves using EEG to characterize the depth of anesthesia through two frequency analysis techniques. On the basis of our analyses, we conclude that the performance of IMF1 is satisfactory in distinguishing between patients' states of consciousness during surgery requiring general anesthesia.

Characterizing Awake and Anesthetized States Using a Dimensionality Reduction Method

Distinguishing between awake and anesthetized states is one of the important problems in surgery. Vital signals contain valuable information that can be used in prediction of different levels of anesthesia. Some monitors based on electroencephalogram (EEG) such as the Bispectral (BIS) index have been proposed in recent years. This study proposes a new method for characterizing between awake and anesthetized states. We validated our method by obtaining data from 25 patients during the cardiac surgery that requires cardiopulmonary bypass. At first, some linear and non-linear features are extracted from EEG signals. Then a method called "LLE"(Locally Linear Embedding) is used to map high-dimensional features in a three-dimensional output space. Finally, low dimensional data are used as an input to a quadratic discriminant analyzer (QDA). The experimental results indicate that an overall accuracy of 88.4 % can be obtained using this method for classifying the EEG signal into conscious and unconscious states for all patients. Considering the reliability of this method, we can develop a new EEG monitoring system that could assist the anesthesiologists to estimate the depth of anesthesia accurately.

Frontal-temporal synchronization of EEG signals quantified by order patterns cross recurrence analysis during propofol anesthesia

Characterizing brain dynamics during anesthesia is a main current challenge in anesthesia study. Several single channel electroencephalogram (EEG)-based commercial monitors like the Bispectral index (BIS) have suggested to examine EEG signal. But, the BIS index has obtained numerous critiques. In this study, we evaluate the concentration-dependent effect of the propofol on long-range frontal-temporal synchronization of EEG signals collected from eight subjects during a controlled induction and recovery design. We used order patterns cross recurrence plot and provide an index named order pattern laminarity (OPL) to assess changes in neuronal synchronization as the mechanism forming the foundation of conscious perception. The prediction probability of 0.9 and 0.84 for OPL and BIS specified that the OPL index correlated more strongly with effect-site propofol concentration. Also, our new index makes faster reaction to transients in EEG recordings based on pharmacokinetic and pharmacodynamic model parameters and demonstrates less variability at the point of loss of consciousness (standard deviation of 0.04 for OPL compared with 0.09 for BIS index). The result show that the OPL index can estimate anesthetic state of patient more efficiently than the BIS index in lightly sedated state with more tolerant of artifacts.

Monitoring depth of anesthesia using combination of EEG measure and hemodynamic variables

Monitoring depth of anesthesia (DOA) via vital signs is a major ongoing challenge for anesthetists. A number of electroencephalogram (EEG)-based monitors such as the Bispectral (BIS) index have been proposed. However, anesthesia is related to central and autonomic nervous system functions whereas the EEG signal originates only from the central nervous system. This paper proposes an automated DOA detection system which consists of three steps. Initially, we introduce multiscale modified permutation entropy index which is robust in the characterization of the burst suppression pattern and combine multiscale information. This index quantifies the amount of complexity in EEG data and is computationally efficient, conceptually simple and artifact resistant. Then, autonomic nervous system activity is quantified with heart rate and mean arterial pressure which are easily acquired using routine monitoring machine. Finally, the extracted features are used as input to a linear discriminate analyzer (LDA). The method is validated with data obtained from 25 patients during the cardiac surgery requiring cardiopulmonary bypass. The experimental results indicate that an overall accuracy of 89.4 % can be obtained using combination of EEG measure and hemodynamic variables, together with LDA to classify the vital sign into awake, light, surgical and deep anesthetised states. The results demonstrate that the proposed method can estimate DOA more effectively than the commercial BIS index with a stronger artifact-resistance.

Monitoring the depth of anesthesia using entropy features and an artificial neural network

Monitoring the depth of anesthesia using an electroencephalogram (EEG) is a major ongoing challenge for anesthetists. The EEG is a recording of brain electrical activity, and it contains valuable information related to the different physiological states of the brain. This study proposes a novel automated method consisting of two steps for assessing anesthesia depth. Initially, the sample entropy and permutation entropy features were extracted from the EEG signal. Because EEG-derived parameters represent different aspects of the EEG features, it would be reasonable to use multiple parameters to assess the effect of the anesthetic. The sample entropy and permutation entropy features quantified the amount of complexity or irregularity in the EEG data and were conceptually simple, computationally efficient and artifact-resistant. Next, the extracted features were used as input for an artificial neural network, which was a data processing system based on the structure of a biological nervous system. The experimental results indicated that an overall accuracy of 88% could be obtained during sevoflurane anesthesia in 17 patients to classify the EEG data into awake, light, general and deep anesthetized states. In addition, this method yielded a classification accuracy of 92.4% to distinguish between awake and general anesthesia in an independent database of propofol and desflurane anesthesia in 129 patients. Considering the high accuracy of this method, a new EEG monitoring system could be developed to assist the anesthesiologist in estimating the depth of anesthesia in a rapid and accurate manner.

Measuring the effects of sevoflurane on electroencephalogram using sample entropy

BACKGROUND

Monitoring the effect of anesthetic drugs on the neural system is a major ongoing challenge for anesthetists. During the past few years, several electroencephalogram (EEG)-based methods such as the response entropy (RE) as implemented in the Datex-Ohmeda M-Entropy Module have been proposed. In this paper, sample entropy is used to quantify the predictability of EEG series, which could provide an index to show the effect of sevoflurane anesthesia. The dose-response relation of sample entropy is compared with that of RE.

METHODS

EEG data from 21 subjects is collected during the induction of general anesthesia with sevoflurane. The sample entropy is applied to the EEG recording. Pharmacokinetic-pharmacodynamic modeling and prediction probability statistic are used to evaluate the efficiency of sample entropy in comparison with RE.

RESULTS

Both methods track the gross changes in EEG, especially the occurrence of burst-suppression pattern at high doses of anesthetics. However, our method produces faster reaction to transients in EEG during the induction of anesthesia as indicated from the pharmacokinetic and pharmacodynamic modeled parameters and analysis around the point of loss of consciousness. Also, sample entropy correlated more closely with effect-site sevoflurane concentration than the RE. In addition, our proposed method exhibits greater resistance to noise in the EEG signals.

CONCLUSION

The results demonstrate that sample entropy can estimate the sevoflurane drug effect on the EEG more effectively than the commercial RE index with a stronger noise resistance.