Data Driven Investigation of Bispectral Index Algorithm

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.

The effect of BIS-guided anaesthesia on the incidence of postoperative nausea and vomiting in children: a prospective randomized double-blind study

BACKGROUND

Postoperative nausea and vomiting (PONV) is a significant problem following paediatric surgery, and volatile anaesthetics are an important cause of this phenomenon. BIS-guided anaesthesia, by reducing the consumption of anaesthetics, leads to a decrease in PONV in adult patients.

STUDY OBJECTIVE

Evaluate the role of BIS-guided anaesthesia in reducing the incidence of paediatric PONV.

DESIGN

Prospective, randomized, double-blind study.

SETTING

A single center study in university hospital in Czech republic, from June 2021 to November 2022.

PATIENTS

A total of 163 children, aged 3-8 years with ASA I-II who underwent endoscopic adenoidectomy under general anaesthesia were included.

INTERVENTIONS

In the intervention group, the depth of anaesthesia was maintained to values between 40 and 60 of BIS.

MAIN OUTCOME MEASURE

The primary outcome was the incidence of postoperative nausea and vomiting during 24 h after surgery.

RESULTS

The use of BIS-guided anaesthesia led to a significant decrease in the incidence of nausea and vomiting compared to the control group [17% vs. 53%; RR (95%CI) 0.48 (0.27-0.86); p < 0.001and 16% vs. 34%; RR (95%CI) 0.33 (0.20-0.54); p = 0.01, respectively].

CONCLUSIONS

BIS-guided anaesthesia decreases the incidence of postoperative nausea and vomiting in children undergoing adenoidectomy.

TRIAL REGISTRATION

Clinicaltrials.gov identifier: NCT04466579.

Development of a Bispectral index score prediction model based on an interpretable deep learning algorithm

BACKGROUND

Proper maintenance of hypnosis is crucial for ensuring the safety of patients undergoing surgery. Accordingly, indicators, such as the Bispectral index (BIS), have been developed to monitor hypnotic levels. However, the black-box nature of the algorithm coupled with the hardware makes it challenging to understand the underlying mechanisms of the algorithms and integrate them with other monitoring systems, thereby limiting their use.

OBJECTIVE

We propose an interpretable deep learning model that forecasts BIS values 25 s in advance using 30 s electroencephalogram (EEG) data.

MATERIAL AND METHODS

The proposed model utilized EEG data as a predictor, which is then decomposed into amplitude and phase components using fast Fourier Transform. An attention mechanism was applied to interpret the importance of these components in predicting BIS. The predictability of the model was evaluated on both regression and binary classification tasks, where the former involved predicting a continuous BIS value, and the latter involved classifying a dichotomous status at a BIS value of 60. To evaluate the interpretability of the model, we analyzed the attention values expressed in the amplitude and phase components according to five ranges of BIS values. The proposed model was trained and evaluated using datasets collected from two separate medical institutions.

RESULTS AND CONCLUSION

The proposed model achieved excellent performance on both the internal and external validation datasets. The model achieved a root-mean-square error of 6.614 for the regression task, and an area under the receiver operating characteristic curve of 0.937 for the binary classification task. Interpretability analysis provided insight into the relationship between EEG frequency components and BIS values. Specifically, the attention mechanism revealed that higher BIS values were associated with increased amplitude attention values in high-frequency bands and increased phase attention values in various frequency bands. This finding is expected to facilitate a more profound understanding of the BIS prediction mechanism, thereby contributing to the advancement of anesthesia technologies.

The Current of Consciousness: Neural Correlates and Clinical Aspects

PURPOSE OF REVIEW

In this review, we summarize the current understanding of consciousness including its neuroanatomic basis. We discuss major theories of consciousness, physical exam-based and electroencephalographic metrics used to stratify levels of consciousness, and tools used to shed light on the neural correlates of the conscious experience. Lastly, we review an expanded category of 'disorders of consciousness,' which includes disorders that impact either the level or experience of consciousness.

RECENT FINDINGS

Recent studies have revealed many of the requisite EEG, ERP, and fMRI signals to predict aspects of the conscious experience. Neurological disorders that disrupt the reticular activating system can affect the level of consciousness, whereas cortical disorders from seizures and migraines to strokes and dementia may disrupt phenomenal consciousness. The recently introduced memory theory of consciousness provides a new explanation of phenomenal consciousness that may explain better than prior theories both experimental studies and the neurologist's clinical experience. Although the complete neurobiological basis of consciousness remains a mystery, recent advances have improved our understanding of the physiology underlying level of consciousness and phenomenal consciousness.

Neuro-functional modeling of near-death experiences in contexts of altered states of consciousness

Near-death experiences (NDEs) including out-of-body experiences (OBEs) have been fascinating phenomena of perception both for affected persons and for communities in science and medicine. Modern progress in the recording of changing brain functions during the time between clinical death and brain death opened the perspective to address and understand the generation of NDEs in brain states of altered consciousness. Changes of consciousness can experimentally be induced in well-controlled clinical or laboratory settings. Reports of the persons having experienced the changes can inform about the similarity of the experiences with those from original NDEs. Thus, we collected neuro-functional models of NDEs including OBEs with experimental backgrounds of drug consumption, epilepsy, brain stimulation, and ischemic stress, and included so far largely unappreciated data from fighter pilot tests under gravitational stress generating cephalic nervous system ischemia. Since we found a large overlap of NDE themes or topics from original NDE reports with those from neuro-functional NDE models, we can state that, collectively, the models offer scientifically appropriate causal explanations for the occurrence of NDEs. The generation of OBEs, one of the NDE themes, can be localized in the temporo-parietal junction (TPJ) of the brain, a multimodal association area. The evaluated literature suggests that NDEs may emerge as hallucination-like phenomena from a brain in altered states of consciousness (ASCs).

Neural complexity is a common denominator of human consciousness across diverse regimes of cortical dynamics

What is the common denominator of consciousness across divergent regimes of cortical dynamics? Does consciousness show itself in decibels or in bits? To address these questions, we introduce a testbed for evaluating electroencephalogram (EEG) biomarkers of consciousness using dissociations between neural oscillations and consciousness caused by rare genetic disorders. Children with Angelman syndrome (AS) exhibit sleep-like neural dynamics during wakefulness. Conversely, children with duplication 15q11.2-13.1 syndrome (Dup15q) exhibit wake-like neural dynamics during non-rapid eye movement (NREM) sleep. To identify highly generalizable biomarkers of consciousness, we trained regularized logistic regression classifiers on EEG data from wakefulness and NREM sleep in children with AS using both entropy measures of neural complexity and spectral (i.e., neural oscillatory) EEG features. For each set of features, we then validated these classifiers using EEG from neurotypical (NT) children and abnormal EEGs from children with Dup15q. Our results show that the classification performance of entropy-based EEG biomarkers of conscious state is not upper-bounded by that of spectral EEG features, which are outperformed by entropy features. Entropy-based biomarkers of consciousness may thus be highly adaptable and should be investigated further in situations where spectral EEG features have shown limited success, such as detecting covert consciousness or anesthesia awareness.

Open Reimplementation of the BIS Algorithms for Depth of Anesthesia

BACKGROUND

BIS (a brand of processed electroencephalogram [EEG] depth-of-anesthesia monitor) scores have become interwoven into clinical anesthesia care and research. Yet, the algorithms used by such monitors remain proprietary. We do not actually know what we are measuring. If we knew, we could better understand the clinical prognostic significance of deviations in the score and make greater research advances in closed-loop control or avoiding postoperative cognitive dysfunction or juvenile neurological injury. In previous work, an A-2000 BIS monitor was forensically disassembled and its algorithms (the BIS Engine) retrieved as machine code. Development of an emulator allowed BIS scores to be calculated from arbitrary EEG data for the first time. We now address the fundamental questions of how these algorithms function and what they represent physiologically.

METHODS

EEG data were obtained during induction, maintenance, and emergence from 12 patients receiving customary anesthetic management for orthopedic, general, vascular, and neurosurgical procedures. These data were used to trigger the closely monitored execution of the various parts of the BIS Engine, allowing it to be reimplemented in a high-level language as an algorithm entitled ibis. Ibis was then rewritten for concision and physiological clarity to produce a novel completely clear-box depth-of-anesthesia algorithm titled openibis .

RESULTS

The output of the ibis algorithm is functionally indistinguishable from the native BIS A-2000, with r = 0.9970 (0.9970-0.9971) and Bland-Altman mean difference between methods of -0.25 ± 2.6 on a unitless 0 to 100 depth-of-anesthesia scale. This precision exceeds the performance of any earlier attempt to reimplement the function of the BIS algorithms. The openibis algorithm also matches the output of the native algorithm very closely ( r = 0.9395 [0.9390-0.9400], Bland-Altman 2.62 ± 12.0) in only 64 lines of readable code whose function can be unambiguously related to observable features in the EEG signal. The operation of the openibis algorithm is described in an intuitive, graphical form.

CONCLUSIONS

The openibis algorithm finally provides definitive answers about the BIS: the reliance of the most important signal components on the low-gamma waveband and how these components are weighted against each other. Reverse engineering allows these conclusions to be reached with a clarity and precision that cannot be obtained by other means. These results contradict previous review articles that were believed to be authoritative: the BIS score does not appear to depend on a bispectral index at all. These results put clinical anesthesia research using depth-of-anesthesia scores on a firm footing by elucidating their physiological basis and enabling comparison to other animal models for mechanistic research.

Spectral edge frequency during general anaesthesia: A narrative literature review

Previous studies have attempted to determine the depth of anaesthesia with different anaesthetic agents using electroencephalogram (EEG) measurements with variable success. Measuring depth of anaesthesia is confounded by the complexity of the EEG and the fact that different agents create different pattens. A narrative review was undertaken to examine the available research evidence on the effect and reliability of spectral edge frequency (SEF) for assessing the depth of anaesthesia in adult patients under general anaesthesia. A systematic search of the PubMed®, Scopus®, CINAHL and Cochrane databases identified six randomized controlled trials and five observational studies. The findings of these studies suggest that SEF varies according to the anaesthetic drugs used. Remifentanil and age are two factors that can affect SEF, while other opioids and benzodiazepine (administered separately) seem to have no effect. No patients experienced intraoperative awareness. However, this does not indicate that SEF can provide full protection against it and the number of articles in which intraoperative awareness was studied was too small to afford any certainty. None of the studies demonstrated a reliable SEF interval associated with adequate general anaesthesia. SEF must be adapted to the anaesthetic drug used, the patient’s age and state while under general anaesthesia.

Usefulness of bispectral index and patient state index during sevoflurane anesthesia in children: A prospective observational study

Cortical electroencephalography-based devices are used to monitor the depth of anesthesia. In this study, we evaluated the values of bispectral index (BIS) and patient state index (PSI) during sevoflurane anesthesia in children. The ability/accuracy of BIS and PSI to predict the maintenance and recovery state of anesthesia was evaluated based on prediction probability (Pk) values and the secondary outcomes were agreement and correlation of 2 monitors. Fifty children (3-12 years old) were enrolled and the patients received sevoflurane anesthesia with remifentanil followed by propofol administration. Before the induction of anesthesia, BIS and PSI sensors were simultaneously placed on the forehead, and data were collected until the end of anesthesia. Maintenance state was defined as the period following intubation until the cessation of sevoflurane, while recovery state was defined as the period following the cessation of sevoflurane until awake. Pk, agreement or correlation of BIS and PSI in different anesthesia state were calculated. Anesthesia reduced mean BIS and PSI values. Pk of BIS (95% confidential interval [CI]: 0.78-0.91) and PSI (95% CI: 0.82-0.91) for anesthesia were 0.85 and 0.87, respectively. Agreement was 0.79 for recovery state and 0.73 for maintenance state. Pk values were comparable for BIS and PSI. Agreement between BIS and PSI measurements in the same state was relatively good. Therefore, these monitors are appropriate for monitoring for different state of anesthesia in pediatric population.

VitalDB, a high-fidelity multi-parameter vital signs database in surgical patients

In modern anesthesia, multiple medical devices are used simultaneously to comprehensively monitor real-time vital signs to optimize patient care and improve surgical outcomes. However, interpreting the dynamic changes of time-series biosignals and their correlations is a difficult task even for experienced anesthesiologists. Recent advanced machine learning technologies have shown promising results in biosignal analysis, however, research and development in this area is relatively slow due to the lack of biosignal datasets for machine learning. The VitalDB (Vital Signs DataBase) is an open dataset created specifically to facilitate machine learning studies related to monitoring vital signs in surgical patients. This dataset contains high-resolution multi-parameter data from 6,388 cases, including 486,451 waveform and numeric data tracks of 196 intraoperative monitoring parameters, 73 perioperative clinical parameters, and 34 time-series laboratory result parameters. All data is stored in the public cloud after anonymization. The dataset can be freely accessed and analysed using application programming interfaces and Python library. The VitalDB public dataset is expected to be a valuable resource for biosignal research and development.

Measurement(s)	vital signs of patients during surgery • perioperative patient information
Technology Type(s)	Vital Signs Measurement • Electronic Medical Record
Factor Type(s)	vital signs data including various numeric and waveform data acquired from multiple patient monitors • perioperative patient information acquired from the electronic medical record system
Sample Characteristic - Organism	Homo sapiens
Sample Characteristic - Environment	hospital
Sample Characteristic - Location	South Korea

Effect of Carbon Dioxide on Bispectral Index of EEG under Intravenous Target-Controlled Anesthesia Based on Intelligent Medical Treatment

Laparoscopic surgery has the advantages of less trauma and quick recovery, and it is more and more favored by surgeons and patients in clinical practice. However, the impact of carbon dioxide pneumoperitoneum on the body during laparoscopic surgery has attracted the attention of many scholars. Pneumoperitoneum can cause increased cerebral blood flow and increased intracranial pressure, cerebral metabolic rate is highly correlated with blood carbon dioxide partial pressure, and cerebral metabolism without cardiopulmonary bypass is linearly correlated with the depth of anesthesia. Electroencephalographic (EEG) bispectral index (BIS) is a signal analysis method, which can directly measure the effect of drugs on the cerebral cortex and reflect the depth of anesthesia. Based on this, this study takes smart medical treatment as the background and uses the improved BP neural network as a tool to explore the effect of carbon dioxide on EEG bispectral index under intravenous target-controlled anesthesia. The main purpose is to observe the correlation between arterial blood carbon dioxide partial pressure and EEG bispectral index under propofol target-controlled anesthesia during retroperitoneal laparoscopic surgery. The experimental results show that the model proposed in this study can efficiently and accurately obtain the size of the influencing factors, which provides a clinical basis for the anesthesia management and anesthesia depth regulation of carbon dioxide pneumoperitoneum laparoscopic surgery.

A Combinatorial Deep Learning Structure for Precise Depth of Anesthesia Estimation From EEG Signals

Electroencephalography (EEG) is commonly used to measure the depth of anesthesia (DOA) because EEG reflects surgical pain and state of the brain. However, precise and real-time estimation of DOA index for painful surgical operations is challenging due to problems such as postoperative complications and accidental awareness. To tackle these problems, we propose a new combinatorial deep learning structure involving convolutional neural networks (inspired by the inception module), bidirectional long short-term memory, and an attention layer. The proposed model uses the EEG signal to continuously predicts the bispectral index (BIS). It is trained over a large dataset, mostly from those under general anesthesia with few cases receiving sedation/analgesia and spinal anesthesia. Despite the imbalance distribution of BIS values in different levels of anesthesia, our proposed structure achieves convincing root mean square error of 5.59 ± 1.04 and mean absolute error of 4.3 ± 0.87, as well as improvement in area under the curve of 15% on average, which surpasses state-of-the-art DOA estimation methods. The DOA values are also discretized into four levels of anesthesia and the results demonstrate strong inter-subject classification accuracy of 88.7% that outperforms the conventional methods.

Depth of Anesthesia Monitoring

The electroencephalogram (EEG) can be analyzed in its raw form for characteristic drug-induced patterns of change or summarized using mathematical parameters as a processed electroencephalogram (pEEG). In this article we aim to summarize the contemporary literature pertaining to the commonly available pEEG monitors including the effects of commonly used anesthetic drugs on the EEG and pEEG parameters, pEEG monitor pitfalls, and the clinical implications of pEEG monitoring for anesthesia, pediatrics, and intensive care.

Heart rate variability-derived features based on deep neural network for distinguishing different anaesthesia states

Background

Estimating the depth of anaesthesia (DoA) is critical in modern anaesthetic practice. Multiple DoA monitors based on electroencephalograms (EEGs) have been widely used for DoA monitoring; however, these monitors may be inaccurate under certain conditions. In this work, we hypothesize that heart rate variability (HRV)-derived features based on a deep neural network can distinguish different anaesthesia states, providing a secondary tool for DoA assessment.

Methods

A novel method of distinguishing different anaesthesia states was developed based on four HRV-derived features in the time and frequency domain combined with a deep neural network. Four features were extracted from an electrocardiogram, including the HRV high-frequency power, low-frequency power, high-to-low-frequency power ratio, and sample entropy. Next, these features were used as inputs for the deep neural network, which utilized the expert assessment of consciousness level as the reference output. Finally, the deep neural network was compared with the logistic regression, support vector machine, and decision tree models. The datasets of 23 anaesthesia patients were used to assess the proposed method.

Results

The accuracies of the four models, in distinguishing the anaesthesia states, were 86.2% (logistic regression), 87.5% (support vector machine), 87.2% (decision tree), and 90.1% (deep neural network). The accuracy of deep neural network was higher than those of the logistic regression (p < 0.05), support vector machine (p < 0.05), and decision tree (p < 0.05) approaches. Our method outperformed the logistic regression, support vector machine, and decision tree methods.

Conclusions

The incorporation of four HRV-derived features in the time and frequency domain and a deep neural network could accurately distinguish between different anaesthesia states; however, this study is a pilot feasibility study. The proposed method—with other evaluation methods, such as EEG—is expected to assist anaesthesiologists in the accurate evaluation of the DoA.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12871-021-01285-x.

Poincaré Plot Area of Gamma-Band EEG as a Measure of Emergence From Inhalational General Anesthesia

The Poincaré plot obtained from electroencephalography (EEG) has been used to evaluate the depth of anesthesia. A standalone EEG Analyzer application was developed; raw EEG signals obtained from a bispectral index (BIS) monitor were analyzed using an on-line monitoring system. Correlations between Poincaré plot parameters and other measurements associated with anesthesia depth were evaluated during emergence from inhalational general anesthesia. Of the participants, 20 were adults anesthetized with sevoflurane (adult_{_SEV}), 20 were adults anesthetized with desflurane (adult_{_DES}), and 20 were pediatric patients anesthetized with sevoflurane (ped_{_SEV}). EEG signals were preprocessed through six bandpass digital filters (f0: 0.5–47 Hz, f1: 0.5–8 Hz, f2: 8–13 Hz, f3: 13–20 Hz, f4: 20–30 Hz, and f5: 30–47 Hz). The Poincaré plot-area ratio (PP_AR = PP_{A_fx}/PP_{A_f0}, fx = f1∼f5) was analyzed at five frequency ranges. Regardless of the inhalational anesthetic used, there were strong linear correlations between the logarithm of PP_AR at f5 and BIS (R² = 0.67, 0.79, and 0.71, in the adult_{_SEV}, adult_{_DES}, and ped_{_SEV} groups, respectively). As an additional observation, a part of EMG activity at the gamma range of 30–47 Hz probably influenced the calculations of BIS and PP_{AR_f5} with a non-negligible level. The logarithm of PP_AR in the gamma band was most sensitive to state changes during the emergence process and could provide a new non-proprietary parameter that correlates with changes in BIS during measurement of anesthesia depth.

A Forensic Disassembly of the BIS Monitor

BACKGROUND

The bispectral index (BIS) monitor has been available for clinical use for >20 years and has had an immense impact on academic activity in Anesthesiology, with >3000 articles referencing the bispectral index. Despite attempts to infer its algorithms by external observation, its operation has nevertheless remained undescribed, in contrast to the algorithms of other less commercially successful monitors of electroencephalogram (EEG) activity under anesthesia. With the expiration of certain key patents, the time is therefore ripe to examine the operation of the monitor on its own terms through careful dismantling, followed by extraction and examination of its internal software.

METHODS

An A-2000 BIS Monitor (gunmetal blue case, amber monochrome display) was purchased on the secondary market. After identifying the major data processing and storage components, a set of free or inexpensive tools was used to retrieve and disassemble the monitor's onboard software. The software executes primarily on an ARMv7 microprocessor (Sharp/NXP LH77790B) and a digital signal processor (Texas Instruments TMS320C32). The device software can be retrieved directly from the monitor's hardware by using debugging interfaces that have remained in place from its original development.

RESULTS

Critical numerical parameters such as the spectral edge frequency (SEF), total power, and BIS values were retraced from external delivery at the device's serial port back to the point of their calculation in the extracted software. In doing so, the locations of the critical algorithms were determined. To demonstrate the validity of the technique, the algorithms for SEF and total power were disassembled, comprehensively annotated and compared to their theoretically ideal behaviors. A bug was identified in the device's implementation of the SEF algorithm, which can be provoked by a perfectly isoelectric EEG.

CONCLUSIONS

This article demonstrates that the electronic design of the A-2000 BIS Monitor does not pose any insuperable obstacles to retrieving its device software in hexadecimal machine code form directly from the motherboard. This software can be reverse engineered through disassembly and decompilation to reveal the methods by which the BIS monitor implements its algorithms, which ultimately must form the definitive statement of its function. Without further revealing any algorithms that might be considered trade secrets, the manufacturer of the BIS monitor should be encouraged to release the device software in its original format to place BIS-related academic literature on a firm theoretical foundation and to promote further academic development of EEG monitoring algorithms.