EEG Based Monitoring of General Anesthesia: Taking the Next Steps

Processed electroencephalography-guided general anesthesia and norepinephrine requirements: A randomized trial in patients having vascular surgery

STUDY OBJECTIVE

Processed electroencephalography (pEEG) may help clinicians optimize depth of general anesthesia. Avoiding excessive depth of anesthesia may reduce intraoperative hypotension and the need for vasopressors. We tested the hypothesis that pEEG-guided - compared to non-pEEG-guided - general anesthesia reduces the amount of norepinephrine needed to keep intraoperative mean arterial pressure above 65 mmHg in patients having vascular surgery.

DESIGN

Randomized controlled clinical trial.

SETTING

University Medical Center Hamburg-Eppendorf, Hamburg, Germany.

PATIENTS

110 patients having vascular surgery.

INTERVENTIONS

pEEG-guided general anesthesia.

MEASUREMENTS

Our primary endpoint was the average norepinephrine infusion rate from the beginning of induction of anesthesia until the end of surgery.

MAIN RESULT

96 patients were analyzed. The mean ± standard deviation average norepinephrine infusion rate was 0.08 ± 0.04 μg kg-1 min-1 in patients assigned to pEEG-guided and 0.12 ± 0.09 μg kg-1 min-1 in patients assigned to non-pEEG-guided general anesthesia (mean difference 0.04 μg kg-1 min-1, 95% confidence interval 0.01 to 0.07 μg kg-1 min-1, p = 0.004). Patients assigned to pEEG-guided versus non-pEEG-guided general anesthesia, had a median time-weighted minimum alveolar concentration of 0.7 (0.6, 0.8) versus 0.8 (0.7, 0.8) (p = 0.006) and a median percentage of time Patient State Index was <25 of 12 (1, 41) % versus 23 (3, 49) % (p = 0.279).

CONCLUSION

pEEG-guided - compared to non-pEEG-guided - general anesthesia reduced the amount of norepinephrine needed to keep mean arterial pressure above 65 mmHg by about a third in patients having vascular surgery. Whether reduced intraoperative norepinephrine requirements resulting from pEEG-guided general anesthesia translate into improved patient-centered outcomes remains to be determined in larger trials.

Nociception Effect on Frontal Electroencephalogram Waveform and Phase-Amplitude Coupling in Laparoscopic Surgery

BACKGROUND

Electroencephalographic pattern changes during anesthesia reflect the nociception-analgesia balance. Alpha dropout, delta arousal, and beta arousal with noxious stimulation have been described during anesthesia; however, data on the reaction of other electroencephalogram signatures toward nociception are scarce. Analyzing the effects of nociception on different electroencephalogram signatures may help us find new nociception markers in anesthesia and understand the neurophysiology of pain in the brain. This study aimed to analyze the electroencephalographic frequency pattern and phase-amplitude coupling change during laparoscopic surgeries.

METHODS

This study evaluated 34 patients who underwent laparoscopic surgery. The electroencephalogram frequency band power and phase-amplitude coupling of different frequencies were analyzed across 3 stages of laparoscopy: incision, insufflation, and opioid stages. Repeated-measures analysis of variance with a mixed model and the Bonferroni method for multiple comparisons were used to analyze the changes in the electroencephalogram signatures between the preincision and postincision/postinsufflation/postopioid phases.

RESULTS

During noxious stimulation, the frequency spectrum showed obvious decreases in the alpha power percentage after the incision (mean ± standard error of the mean [SEM], 26.27 ± 0.44 and 24.37 ± 0.66; P < .001) and insufflation stages (26.27 ± 0.44 and 24.40 ± 0.68; P = .002), which recovered after opioid administration. Further phase-amplitude analyses showed that the modulation index (MI) of the delta-alpha coupling decreased after the incision stage (1.83 ± 0.22 and 0.98 ± 0.14 [MI × 10 3 ]; P < .001), continued to be suppressed during the insufflation stage (1.83 ± 0.22 and 1.17 ± 0.15 [MI × 10 3 ]; P = .044), and recovered after opioid administration.

CONCLUSIONS

Alpha dropout during noxious stimulation is observed in laparoscopic surgeries under sevoflurane. In addition, the modulation index of delta-alpha coupling decreases during noxious stimulation and recovers after the administration of rescue opioids. Phase-amplitude coupling of the electroencephalogram may be a new approach for evaluating the nociception-analgesia balance during anesthesia.

Electroencephalographic monitoring of anesthesia during surgical procedures in mice using a modified clinical monitoring system

Monitoring brain activity and associated physiology during the administration of general anesthesia (GA) in mice is pivotal to guarantee postanesthetic health. Clinically, electroencephalogram (EEG) monitoring is a well-established method to guide GA. There are no established methods available for monitoring EEG in mice (Mus musculus) during surgery. In this study, a minimally invasive rodent intraoperative EEG monitoring system was implemented using subdermal needle electrodes and a modified EEG-based commercial patient monitor. EEG recordings were acquired at three different isoflurane concentrations revealing that surgical concentrations of isoflurane anesthesia predominantly contained burst suppression patterns in mice. EEG suppression ratios and suppression durations showed strong positive correlations with the isoflurane concentrations. The electroencephalographic indices provided by the monitor did not support online monitoring of the anesthetic status. The online available suppression duration in the raw EEG signals during isoflurane anesthesia is a straight forward and reliable marker to assure safe, adequate and reproducible anesthesia protocols.

Separation of responsive and unresponsive patients under clinical conditions: comparison of symbolic transfer entropy and permutation entropy

Electroencephalogram (EEG)-based monitoring during general anesthesia may help prevent harmful effects of high or low doses of general anesthetics. There is currently no convincing evidence in this regard for the proprietary algorithms of commercially available monitors. The purpose of this study was to investigate whether a more mechanism-based parameter of EEG analysis (symbolic transfer entropy, STE) can separate responsive from unresponsive patients better than a strictly probabilistic parameter (permutation entropy, PE) under clinical conditions. In this prospective single-center study, the EEG of 60 surgical ASA I-III patients was recorded perioperatively. During induction of and emergence from anesthesia, patients were asked to squeeze the investigators' hand every 15s. Time of loss of responsiveness (LoR) during induction and return of responsiveness (RoR) during emergence from anesthesia were registered. PE and STE were calculated at -15s and +30s of LoR and RoR and their ability to separate responsive from unresponsive patients was evaluated using accuracy statistics. 56 patients were included in the final analysis. STE and PE values decreased during anesthesia induction and increased during emergence. Intra-individual consistency was higher during induction than during emergence. Accuracy values during LoR and RoR were 0.71 (0.62-0.79) and 0.60 (0.51-0.69), respectively for STE and 0.74 (0.66-0.82) and 0.62 (0.53-0.71), respectively for PE. For the combination of LoR and RoR, values were 0.65 (0.59-0.71) for STE and 0.68 (0.62-0.74) for PE. The ability to differentiate between the clinical status of (un)responsiveness did not significantly differ between STE and PE at any time. Mechanism-based EEG analysis did not improve differentiation of responsive from unresponsive patients compared to the probabilistic PE.Trial registration: German Clinical Trials Register ID: DRKS00030562, November 4, 2022, retrospectively registered.

Consciousness and General Anesthesia: Challenges for Measuring the Depth of Anesthesia

The optimal consciousness level required for general anesthesia with surgery is unclear, but in existing practice, anesthetic oblivion, may be incomplete. This article discusses the concept of consciousness, how it is altered by anesthetics, the challenges for assessing consciousness, currently used technologies for assessing anesthesia levels, and future research directions. Wakefulness is marked by a subjective experience of existence (consciousness), perception of input from the body or the environment (connectedness), the ability for volitional responsiveness, and a sense of continuity in time. Anesthetic drugs may selectively impair some of these components without complete extinction of the subjective experience of existence. In agreement with Sanders et al. (2012), the authors propose that a state of disconnected consciousness is the optimal level of anesthesia, as it likely avoids both awareness and the possible dangers of oversedation. However, at present, there are no reliably tested indices that can discriminate between connected consciousness, disconnected consciousness, and complete unconsciousness.

Response of the GE Entropy™ monitor to neuromuscular block in awake volunteers

BACKGROUND

The GE Entropy™ monitor analyses the frontal electroencephalogram (EEG) and generates two indices intended to represent the degree of anaesthetic drug effect on the brain. It is frequently used in the context of neuromuscular block. We have shown that a similar device, the Bispectral Index monitor (BIS), does not generate correct values in awake volunteers when neuromuscular blocking drugs are administered.

METHODS

We replayed the EEGs recorded during awake paralysis from the original study to an Entropy monitor via a calibrated electronic playback system. Each EEG was replayed 30 times to evaluate the consistency of the Entropy output.

RESULTS

Both State Entropy and Response Entropy decreased during periods of neuromuscular block to values consistent with anaesthesia, despite there being no change in conscious state (State Entropy <60 in eight of nine rocuronium trials and nine of 10 suxamethonium trials). Entropy values did not return to pre-test levels until after the return of movement. Entropy did not generate exactly the same results when the same EEG was replayed multiple times, which is primarily because of a cyclical state within the Entropy system itself.

CONCLUSIONS

The GE Entropy™ monitor requires muscle activity to generate correct values in an awake subject. It could therefore be unreliable at detecting awareness in patients who have been given neuromuscular blocking drugs. In addition, Entropy does not generate the same result each time it is presented with the same EEG.

Enhanced recovery protocols for ambulatory surgery

INTRODUCTION

In the United States, ambulatory surgeries account for up to 87% of all surgical procedures. (1) It was estimated that 19.2 million ambulatory surgeries were performed in 2018 (https://www.hcup-us.ahrq.gov/reports/statbriefs/sb287-Ambulatory-Surgery-Overview-2019.pdf). Cataract procedures and musculoskeletal surgeries are the most common surgical interventions performed in ambulatory centers. However, more complex surgical interventions, such as sleeve gastrectomies, oncological, and spine surgeries, and even arthroplasties are routinely performed as day cases or in a model of an ambulatory extended recovery. (2-5) The ambulatory surgery centers industry has grown since 2017 by 1.1% per year and reached a market size of $31.2 billion. According to the Ambulatory Surgery Center Association, there is a potential to save $57.6 billion in Medicare costs over the next decade (https://www.ibisworld.com/industry-statistics/market-size/ambulatory-surgery-centers-united-states/). These data suggest an expected rise in the volume of ambulatory (same day) or extended ambulatory (23 h) surgeries in coming years. Similar increases are also observed in other countries. For example, 75% of elective surgeries are performed as same-day surgery in the United Kingdom. (6) To reduce costs and improve the quality of care after those more complex procedures, ambulatory surgery centers have started implementing patient-centered, high-quality, value-based practices. To achieve those goals, Enhanced Recovery After Surgery (ERAS) protocols have been implemented to reduce the length of stay, decrease costs, increase patients' satisfaction, and transform clinical practices. The ERAS fundamentals for ambulatory surgery are based on five pillars, including (1) preoperative patient counseling, education, and optimization; (2) multimodal and opioid-sparing analgesia; (3) nausea and vomiting, wound infection, and venous thromboembolism prophylaxis; (4) maintenance of euvolemia; and (5) encouragement of early mobility. Those pillars rely on interdisciplinary teamwork led by anesthesiologists, surgery-specific workgroups, and safety culture. (2) Research shows that a team of ambulatory anesthesiologists is crucial in improving postoperative nausea and vomiting (PONV) and pain control. (7) This review will summarize the current evidence on the elements and clinical importance of implementing ERAS protocol for ambulatory surgery.

Information-Theoretic Approaches in EEG Correlates of Auditory Perceptual Awareness under Informational Masking

In informational masking paradigms, the successful segregation between the target and masker creates auditory perceptual awareness. The dynamics of the build-up of auditory perception is based on a set of interactions between bottom-up and top-down processes that generate neuronal modifications within the brain network activity. These neural changes are studied here using event-related potentials (ERPs), entropy, and integrated information, leading to several measures applied to electroencephalogram signals. The main findings show that the auditory perceptual awareness stimulated functional activation in the fronto-temporo-parietal brain network through (i) negative temporal and positive centro-parietal ERP components; (ii) an enhanced processing of multi-information in the temporal cortex; and (iii) an increase in informational content in the fronto-central cortex. These different results provide information-based experimental evidence about the functional activation of the fronto-temporo-parietal brain network during auditory perceptual awareness.

A Narrative Review Illustrating the Clinical Utility of Electroencephalogram-Guided Anesthesia Care in Children

The major therapeutic end points of general anesthesia include hypnosis, amnesia, and immobility. There is a complex relationship between general anesthesia, responsiveness, hemodynamic stability, and reaction to noxious stimuli. This complexity is compounded in pediatric anesthesia, where clinicians manage children from a wide range of ages, developmental stages, and body sizes, with their concomitant differences in physiology and pharmacology. This renders anesthetic requirements difficult to predict based solely on a child's age, body weight, and vital signs. Electroencephalogram (EEG) monitoring provides a window into children's brain states and may be useful in guiding clinical anesthesia management. However, many clinicians are unfamiliar with EEG monitoring in children. Young children's EEGs differ substantially from those of older children and adults, and there is a lack of evidence-based guidance on how and when to use the EEG for anesthesia care in children. This narrative review begins by summarizing what is known about EEG monitoring in pediatric anesthesia care. A key knowledge gap in the literature relates to a lack of practical information illustrating the utility of the EEG in clinical management. To address this gap, this narrative review illustrates how the EEG spectrogram can be used to visualize, in real time, brain responses to anesthetic drugs in relation to hemodynamic stability, surgical stimulation, and other interventions such as cardiopulmonary bypass. This review discusses anesthetic management principles in a variety of clinical scenarios, including infants, children with altered conscious levels, children with atypical neurodevelopment, children with hemodynamic instability, children undergoing total intravenous anesthesia, and those undergoing cardiopulmonary bypass. Each scenario is accompanied by practical illustrations of how the EEG can be visualized to help titrate anesthetic dosage to avoid undersedation or oversedation when patients experience hypotension or other physiological challenges, when surgical stimulation increases, and when a child's anesthetic requirements are otherwise less predictable. Overall, this review illustrates how well-established clinical management principles in children can be significantly complemented by the addition of EEG monitoring, thus enabling personalized anesthesia care to enhance patient safety and experience.

EEG responses to standardised noxious stimulation during clinical anaesthesia: a pilot study

Background

During clinical anaesthesia, the administration of analgesics mostly relies on empirical knowledge and observation of the patient's reactions to noxious stimuli. Previous studies in healthy volunteers under controlled conditions revealed EEG activity in response to standardised nociceptive stimuli even at high doses of remifentanil and propofol. This pilot study aims to investigate the feasibility of using these standardised nociceptive stimuli in routine clinical practice.

Methods

We studied 17 patients undergoing orthopaedic trauma surgery under general anaesthesia. We evaluated if the EEG could track standardised noxious phase-locked electrical stimulation and tetanic stimulation, a time-locked surrogate for incisional pain, before, during, and after the induction of general anaesthesia. Subsequently, we analysed the effect of tetanic stimulation on the surgical pleth index as a peripheral, vegetative, nociceptive marker.

Results

We found that the phase-locked evoked potentials after noxious electrical stimulation vanished after the administration of propofol, but not at low concentrations of remifentanil. After noxious tetanic stimulation under general anaesthesia, there were no consistent spectral changes in the EEG, but the vegetative response in the surgical pleth index was statistically significant (Hedges' g effect size 0.32 [95% confidence interval 0.12-0.77], P=0.035).

Conclusion

Our standardised nociceptive stimuli are not optimised for obtaining consistent EEG responses in patients during clinical anaesthesia. To validate and sufficiently reproduce EEG-based standardised stimulation as a marker for nociception in clinical anaesthesia, other pain models or stimulation settings might be required to transfer preclinical studies into clinical practice.

Clinical trial registration

DRKS00017829.

Electroencephalogram-based prediction and detection of responsiveness to noxious stimulation in critical care patients: a retrospective single-centre analysis

BACKGROUND

Monitoring of pain and nociception in critical care patients unable to self-report pain remains a challenge, as clinical signs are neither sensitive nor specific. Available technical approaches are limited by various constraints. We investigated the electroencephalogram (EEG) for correlates that precede or coincide with behavioural nociceptive responses to noxious stimulation.

METHODS

In this retrospective study, we analysed frontal EEG recordings of 64 critical care patients who were tracheally intubated and ventilated before, during, and after tracheal suctioning. We investigated EEG power bands for correlates preceding or coinciding with behavioural responses (Behavioural Pain Scale ≥7). We applied the Mann-Whitney U-test to calculate corresponding P-values.

RESULTS

Strong behavioural responses were preceded by higher normalised power in the 2.5-5 Hz band (+17.1%; P<0.001) and lower normalised power in the 0.1-1.5 Hz band (-10.5%; P=0.029). After the intervention, strong behavioural responses were associated with higher normalised EEG power in the 2.5-5 Hz band (+16.6%; P=0.021) and lower normalised power in the 8-12 Hz band (-51.2%; P=0.037) CONCLUSIONS: We observed correlates in EEG band power that precede and coincide with behavioural responses to noxious stimulation. Based on previous findings, some of the power bands could be linked to processing of nociception, arousal, or sedation effects. The power bands more closely related to nociception and arousal could be used to improve monitoring of nociception and to optimise analgesic management in critical care patients.

CLINICAL TRIAL REGISTRATION

DRKS00011206.

The absence of dominant alpha-oscillatory EEG activity during emergence from delta-dominant anesthesia predicts neurocognitive impairment- results from a prospective observational trial

STUDY OBJECTIVE

Postoperative neurocognitive disorders (PND) are common complications after surgery under general anesthesia. In our aging society the incidence of PND will increase. Hence, interdisciplinary efforts should be taken to minimize the occurrence of PND. Electroencephalographic (EEG) monitoring of brain activity during anesthesia or emergence from anesthesia is a promising tool to identify patients at risk. We therefore investigated whether we could identify specific EEG signatures during emergence of anesthesia that are associated with the occurrence of PND.

DESIGN AND PATIENTS

We performed a prospective observational investigation on 116 patients to evaluate the EEG features during emergence from general anesthesia dominated by slow delta waves in patients with and without delirium in the postoperative care unit (PACU-D) as assessed by the CAM-ICU and the RASS.

MAIN RESULTS

During emergence both the frontal and global EEG of patients with PACU-D were significantly different from patients without PACU-D. PACU-D patients had lower relative alpha power and reduced fronto-parietal alpha coherence.

CONCLUSIONS

With our analysis we show differences in EEG features associated with anesthesia emergence in patients with and without PACU-D. Frontal and global EEG alpha-band features could help to identify patients with PACU-D.

CLINICAL TRIAL NUMBER

NCT03287401.

Neuronal Oscillatory Signatures in the Developing Mouse Visual Cortex After Short-Term Monocular Deprivation

Development and maturation in cortical networks depend on neuronal activity. For stabilization and pruning of connections, synchronized oscillations play a crucial role. A fundamental mechanism that enables coordinated activity during brain functioning is formed of synchronized neuronal oscillations in low- (delta and theta) and high- (gamma) frequency bands. The relationship between neural synchrony, cognition, and the perceptual process has been widely studied, but any possible role of neural synchrony in amblyopia has been less explored. We hypothesized that monocular deprivation (MD) during early postnatal life would lead to changes in neuronal activity that would be demonstrated by changes in phase-amplitude coupling (PAC) and altered power in specific oscillatory frequency. Our results demonstrate that functional connectivity in the visual cortex is altered by MD during adolescence. The amplitude of high-frequency oscillations is modulated by the phase of low-frequency oscillations. Demonstration of enhanced delta–gamma and theta–gamma PAC indicates that our results are relevant for a broad range of nested oscillatory markers. These markers are inherent to neuronal processing and are consistent with the hypothesized increase in the intrinsic coupling that arises from neural oscillatory phase alignment. Our results reveal distinct frequency bands exhibit altered power and coherence variations modulated by experience-driven plasticity.

Robust learning from corrupted EEG with dynamic spatial filtering

Building machine learning models using EEG recorded outside of the laboratory setting requires methods robust to noisy data and randomly missing channels. This need is particularly great when working with sparse EEG montages (1-6 channels), often encountered in consumer-grade or mobile EEG devices. Neither classical machine learning models nor deep neural networks trained end-to-end on EEG are typically designed or tested for robustness to corruption, and especially to randomly missing channels. While some studies have proposed strategies for using data with missing channels, these approaches are not practical when sparse montages are used and computing power is limited (e.g., wearables, cell phones). To tackle this problem, we propose dynamic spatial filtering (DSF), a multi-head attention module that can be plugged in before the first layer of a neural network to handle missing EEG channels by learning to focus on good channels and to ignore bad ones. We tested DSF on public EEG data encompassing ∼4,000 recordings with simulated channel corruption and on a private dataset of ∼100 at-home recordings of mobile EEG with natural corruption. Our proposed approach achieves the same performance as baseline models when no noise is applied, but outperforms baselines by as much as 29.4% accuracy when significant channel corruption is present. Moreover, DSF outputs are interpretable, making it possible to monitor the effective channel importance in real-time. This approach has the potential to enable the analysis of EEG in challenging settings where channel corruption hampers the reading of brain signals.

The Strength of Alpha Oscillations in the Electroencephalogram Differently Affects Algorithms Used for Anesthesia Monitoring

BACKGROUND

Intraoperative patient monitoring using the electroencephalogram (EEG) can help to adequately adjust the anesthetic level. Therefore, the processed EEG (pEEG) provides the anesthesiologist with the estimated anesthesia level. The commonly used approaches track the changes from a fast- and a low-amplitude EEG during wakefulness to a slow- and a high-amplitude EEG under general anesthesia. However, besides these changes, another EEG feature, a strong oscillatory activity in the alpha band (8-12 Hz), develops in the frontal EEG. Strong alpha-band activity during general anesthesia seems to reflect an appropriate anesthetic level for certain anesthetics, but the way the common pEEG approaches react to changes in the alpha-band activity is not well explained. Hence, we investigated the impact of an artificial alpha-band modulation on pEEG approaches used in anesthesia research.

METHODS

We performed our analyses based on 30 seconds of simulated sedation (n = 25) EEG, simulated anesthesia (n = 25) EEG, and EEG episodes from 20 patients extracted from a steady state that showed a clearly identifiable alpha peak in the density spectral array (DSA) and a state entropy (GE Healthcare) around 50, indicative of adequate anesthesia. From these traces, we isolated the alpha activity by band-pass filtering (8-12 Hz) and added this alpha activity to or subtracted it from the signals in a stepwise manner. For each of the original and modified signals, the following pEEG values were calculated: (1) spectral edge frequency (SEF95), (2) beta ratio, (3) spectral entropy (SpEntr), (4) approximate entropy (ApEn), and (5) permutation entropy (PeEn).

RESULTS

The pEEG approaches showed different reactions to the alpha-band modification that depended on the data set and the amplification step. The beta ratio and PeEn decreased with increasing alpha activity for all data sets, indicating a deepening of anesthesia. The other pEEG approaches behaved nonuniformly. SEF95, SpEntr, and ApEn decreased with increasing alpha for the simulated anesthesia data (arousal) but decreased for simulated sedation. For the patient EEG, ApEn indicated an arousal, and SEF95 and SpEntr showed a nonuniform change.

CONCLUSIONS

Changes in the alpha-band activity lead to different reactions for different pEEG approaches. Hence, the presence of strong oscillatory alpha activity that reflects an adequate level of anesthesia may be interpreted differently, by an either increasing (arousal) or decreasing (deepening) pEEG value. This could complicate anesthesia navigation and prevent the adjustment to an adequate, alpha-dominant anesthesia level, when titrating by the pEEG values.

Revisión de electroencefalografía portable y su aplicabilidad en neurociencias

La electroencefalografía (EEG) es una técnica que permite registrar la actividad eléctrica del cerebro y ha sido estudiada durante los últimos cien años en diferentes ámbitos de la neurociencia. En los últimos años se ha investigado y desarrollado equipos de medición que sean portables y que permitan una buena calidad de la señal, por lo cual se realizó una revisión bibliográfica de las compañías fabricantes de algunos dispositivos de electroencefalografía portable disponibles en el mercado, se exponen sus características principales, algunos trabajos encontrados que fueron realizados con los dispositivos, comparaciones entre los mismos y una discusión acerca de las ventajas y desventajas de sus características. Finalmente se concluye que a la hora de comprar un dispositivo para electroencefalografía portable es necesario tener en cuenta el uso que se le va a dar y el costo-beneficio que tiene el equipo de acuerdo con sus características. Encephalography is a technique that allows the recording of electrical activity of the brain and has been studied during the last hundred years in different areas of neuroscience. For several years, measuring equipment that are portable and that allow a good signal quality to have been researched and developed, so a literature review of the manufacturing companies of some of portable electroencephalography devices available on the market was carried out: Its main features are exposed, as well as some of the work found that were made with those, comparisons between them and a discussion about the advantages and disadvantages of their features. It is concluded that, when a portable encephalography device is bought, it’s necessary to take into consideration the use that it will be having and the cost-benefit that the device has according to its features.

Ketamine-xylazine anesthesia depth affects auditory neuronal responses in the lateral superior olive complex of the gerbil

Studies of in vivo neuronal responses to auditory inputs in the superior olive complex (SOC) are usually done under anesthesia. However, little attention has been paid to the effect of anesthesia itself on response properties. Here, we assessed the effect of anesthesia depth under ketamine-xylazine anesthetics on auditory evoked response properties of lateral SOC neurons. Anesthesia depth was tracked by monitoring EEG spectral peak frequencies. An increase in anesthesia depth led to a decrease of spontaneous discharge activities and an elevated response threshold. The temporal responses to suprathreshold tones were also affected, with adapted responses reduced but peak responses unaffected. Deepening the anesthesia depth also increased first spike latency. However, spike jitter was not affected. Auditory brainstem responses to clicks confirmed that ketamine-xylazine anesthesia depth affects auditory neuronal activities and the effect on spike rate and spike timing persists through the auditory pathway. We concluded from those observations that ketamine-xylazine affects lateral SOC response properties depending on the anesthesia depth.NEW & NOTEWORTHY We studied how the depth of ketamine-xylazine anesthesia altered response properties of lateral superior olive complex neurons, and auditory brainstem evoked responses. Our results provide direct evidence that anesthesia depth affects auditory neuronal responses and reinforce the notion that both the anesthetics and the anesthesia depth should be considered when interpreting/comparing in vivo neuronal recordings.

The First Derivative of the Electroencephalogram Facilitates Tracking of Electroencephalographic Alpha Band Activity During General Anesthesia

BACKGROUND

Intraoperative neuromonitoring can help to navigate anesthesia. Pronounced alpha oscillations in the frontal electroencephalogram (EEG) appear to predict favorable perioperative neurocognitive outcomes and may also provide a measure of intraoperative antinociception. Monitoring the presence and strength of these alpha oscillations can be challenging, especially in elderly patients, because the EEG in these patients may be dominated by oscillations in other frequencies. Hence, the information regarding alpha oscillatory activity may be hidden and hard to visualize on a screen. Therefore, we developed an effective approach to improve the detection and presentation of alpha activity in the perioperative setting.

METHODS

We analyzed EEG records of 180 patients with a median age of 60 years (range, 18-90 years) undergoing noncardiac, nonneurologic surgery under general anesthesia with propofol induction and sevoflurane maintenance. We calculated the power spectral density (PSD) for the unprocessed EEG as well as for the time-discrete first derivative of the EEG (diffPSD) from 10-second epochs. Based on these data, we estimated the power-law coefficient κ of the PSD and diffPSD, as the EEG coarsely follows a 1/fκ distribution when displayed in double logarithmic coordinates. In addition, we calculated the alpha (7.8-12.1 Hz) to delta (0.4-4.3 Hz) ratio from the PSD as well as diffPSD.

RESULTS

The median κ was 0.899 [first and third quartile: 0.786, 0.986] for the unaltered PSD, and κ = -0.092 [-0.202, -0.013] for the diffPSD, corresponding to an almost horizontal PSD of the differentiated EEG. The alpha-to-delta ratio of the diffPSD was strongly increased (median ratio = -8.0 dB [-10.5, -4.7 dB] for the unaltered PSD versus 30.1 dB [26.1, 33.8 dB] for the diffPSD). A strong narrowband oscillatory alpha power component (>20% of total alpha power) was detected in 23% using PSD, but in 96% of the diffPSD.

CONCLUSIONS

We demonstrated that the calculation of the diffPSD from the time-discrete derivative of the intraoperative frontal EEG is a straightforward approach to improve the detection of alpha activity by eliminating the broadband background noise. This improvement in alpha peak detection and visualization could facilitate the guidance of general anesthesia and improve patient outcome.

Machine Learning Based Hardware Architecture for DOA Measurement from Mice EEG

OBJECTIVE

This research aims to design a hardware optimized machine learning based Depth of Anesthesia (DOA) measurement framework for mice and its FPGA implementation.

METHODS

Electroencephalography or EEG signal is acquired from 16 mice in the Neural Interface Research (NIR) Laboratory of the City University of Hong Kong. We present a logistic regression based approach with mathematically uncomplicated feature extraction techniques for efficient hardware implementation to estimate the DOA.

RESULTS

With the extraction of only two features, the proposed system can classify the state of consciousness with 94% accuracy for a 1 second EEG epoch, leading to a 100% accurate channel prediction after a 7 second run-time on average.

CONCLUSION

Through performance evaluation and comparative study confirmed the efficacy of the prototype.

SIGNIFICANCE

Traditionally the DOA is estimated by checking biophysical responses of a patient during the surgery. However, the physical symptoms can be misleading for a decisive conclusion due to the patient's health condition or as a side-effect of anesthetic drugs. Recently, several neuroscientific research works are correlating the EEG signal with conscious states, which is likely to have less interference with the patient's medical condition. This research presents the first-of-its-kind hardware implemented automatic DOA computation system for mice.

Effects of noxious stimulation on the electroencephalogram during general anaesthesia: a narrative review and approach to analgesic titration

Electroencephalographic (EEG) activity is used to monitor the neurophysiology of the brain, which is a target organ of general anaesthesia. Besides its use in evaluating hypnotic states, neurophysiologic reactions to noxious stimulation can also be observed in the EEG. Recognising and understanding these responses could help optimise intraoperative analgesic management. This review describes three types of changes in the EEG induced by noxious stimulation when the patient is under general anaesthesia: (1) beta arousal, (2) (paradoxical) delta arousal, and (3) alpha dropout. Beta arousal is an increase in EEG power in the beta-frequency band (12-25 Hz) in response to noxious stimulation, especially at lower doses of anaesthesia drugs in the absence of opioids. It is usually indicative of a cortical depolarisation and increased cortical activity. At higher concentrations of anaesthetic drug, and with insufficient opioids, delta arousal (increased power in the delta band [0.5-4 Hz]) and alpha dropout (decreased alpha power [8-12 Hz]) are associated with noxious stimuli. The mechanisms of delta arousal are not well understood, but the midbrain reticular formation seems to play a role. Alpha dropout may indicate a return of thalamocortical communication, from an idling mode to an operational mode. Each of these EEG changes reflect an incomplete modulation of pain signals and can be mitigated by administration of opioid or the use of regional anaesthesia techniques. Future studies should evaluate whether titrating analgesic drugs in response to these EEG signals reduces postoperative pain and influences other postoperative outcomes, including the potential development of chronic pain.

Frontal electroencephalogram based drug, sex, and age independent sedation level prediction using non-linear machine learning algorithms

Brain monitors which track quantitative electroencephalogram (EEG) signatures to monitor sedation levels are drug and patient specific. There is a need for robust sedation level monitoring systems to accurately track sedation levels across all drug classes, sex and age groups. Forty-four quantitative features estimated from a pooled dataset of 204 EEG recordings from 66 healthy adult volunteers who received either propofol, dexmedetomidine, or sevoflurane (all with and without remifentanil) were used in a machine learning based automated system to estimate the depth of sedation. Model training and evaluation were performed using leave-one-out cross validation methodology. We trained four machine learning models to predict sedation levels and evaluated the influence of remifentanil, age, and sex on the prediction performance. The area under the receiver-operator characteristic curve (AUC) was used to assess the performance of the prediction model. The ensemble tree with bagging outperformed other machine learning models and predicted sedation levels with an AUC = 0.88 (0.81–0.90). There were significant differences in the prediction probability of the automated systems when trained and tested across different age groups and sex. The performance of the EEG based sedation level prediction system is drug, sex, and age specific. Nonlinear machine-learning models using quantitative EEG features can accurately predict sedation levels. The results obtained in this study may provide a useful reference for developing next generation EEG based sedation level prediction systems using advanced machine learning algorithms.

Clinical trial registration: NCT 02043938 and NCT 03143972.

Assessment of changes in the electrical activity of the brain during general anesthesia using portable electroencephalography

Introduction: The analysis of the electrical activity of the brain using scalp electrodes with electroencephalography (EEG) could reveal the depth of anesthesia of a patient during surgery. However, conventional EEG equipment, due to its price and size, are not a practical option for the operating room and the commercial units used in surgery do not provide access to the electrical activity. The availability of low-cost portable technologies could provide for further research on the brain activity under general anesthesia and facilitate our quest for new markers of depth of anesthesia. Objective: To assess the capabilities of a portable EEG technology to capture brain rhythms associated with the state of consciousness and the general anesthesia status of surgical patients anesthetized with propofol. Methods: Observational, cross-sectional trial that reviewed 10 EEG recordings captured using OpenBCI portable low-cost technology, in female patients undergoing general anesthesia with propofol. The signal from the frontal electrodes was analyzed with spectral analysis and the results were compared against the reports in the literature. Results: The signal captured with frontal electrodes, particularly α rhythm, enabled the distinction between resting with eyes closed and with eyes opened in a conscious state, and sustained anesthesia during surgery. Conclusions: It is possible to differentiate a resting state from sustained anesthesia, replicating previous findings with conventional technologies. These results pave the way to the use of portable technologies such as the OpenBCI tool, to explore the brain dynamics during anesthesia.

The influence of induction speed on the frontal (processed) EEG

The intravenous injection of the anaesthetic propofol is clinical routine to induce loss of responsiveness (LOR). However, there are only a few studies investigating the influence of the injection rate on the frontal electroencephalogram (EEG) during LOR. Therefore, we focused on changes of the frontal EEG especially during this period. We included 18 patients which were randomly assigned to a slow or fast induction group and recorded the frontal EEG. Based on this data, we calculated the power spectral density, the band powers and band ratios. To analyse the behaviour of processed EEG parameters we calculated the beta ratio, the spectral entropy, and the spectral edge frequency. Due to the prolonged induction period in the slow injection group we were able to distinguish loss of responsiveness to verbal command (LOvR) from loss of responsiveness to painful stimulus (LOpR) whereas in the fast induction group we could not. At LOpR, we observed a higher relative alpha and beta power in the slow induction group while the relative power in the delta range was lower than in the fast induction group. When concentrating on the slow induction group the increase in relative alpha power pre-LOpR and even before LOvR indicated that frontal EEG patterns, which have been suggested as an indicator of unconsciousness, can develop before LOR. Further, LOvR was best reflected by an increase of the alpha to delta ratio, and LOpR was indicated by a decrease of the beta to alpha ratio. These findings highlight the different spectral properties of the EEG at various levels of responsiveness and underline the influence of the propofol injection rate on the frontal EEG during induction of general anesthesia.

Wavelet and pain rating index for inhalation anesthesia: A randomized controlled trial

BACKGROUND

Wavelet index (WLi) and pain rating index (PRi) are new parameters for regulating general anesthesia depth based on wavelet analysis.

AIM

To investigate the safety and efficacy of using WLi or PRi in sevoflurane anesthesia.

METHODS

This randomized controlled trial enrolled 66 patients scheduled for elective posterior lumbar interbody fusion surgery under sevoflurane anesthesia between September 2017 and February 2018. A random number generator was used to assign the eligible patients to three groups: Systolic blood pressure (SBP) monitoring group, WLi monitoring group, and PRi monitoring group. The main anesthesiologist was aware of the patient grouping and intervention used. The primary endpoint was anesthesia recovery time. Secondary endpoints included extubation time, sevoflurane consumption, number of unwanted events/ interventions, number of adverse events and postoperative visual analogue scale for pain.

RESULTS

A total of 62 patients were included in the final analysis (SBP group, n = 21; WLi group, n = 21; and PRi group, n = 20). There were no significant differences among the three groups in patient age, gender distribution, body mass index, American Society of Anesthesiologists class, duration of surgery, or duration of anesthesia. Anesthesia recovery time was shorter in the WLi and PRi groups than in the SBP group with no significant difference between the WLi and PRi groups. Extubation time was shorter in the WLi and PRi groups than in the SBP group. Sevoflurane consumption was lower in the WLi and PRi groups than in the SBP group. Nicardipine was more commonly needed to treat hypertension in the WLi and PRi groups than in the SBP group.

CONCLUSION

Regulation of sevoflurane anesthesia depth with WLi or PRi reduced anesthesia recovery time, extubation time and sevoflurane consumption without intraoperative unwanted events.

Use of Processed Electroencephalography in the Clinical Setting

Purpose of Review

Processed electroencephalography (pEEG) is widely used in clinical practice. Few clinicians utilize the full potential of these devices. This brief review will address the improvements in patient management available from the utilization of all pEEG data.

Recent Findings

Anesthesiologists easily learn to recognize raw pEEG patterns that are consistent with an appropriate level of hypnotic effect. Power distribution within the waveform can be displayed in a visual format that identifies signatures of the principal anesthetic hypnotics. Opinion on the benefit of pEEG data in the mitigation of postoperative neurological impairment remains divided.

Summary

Looking beyond the index number can aid clinical decision making and improve confidence in the benefits of this monitoring modality.

Electroencephalographic Alpha and Delta Oscillation Dynamics in Response to Increasing Doses of Propofol

BACKGROUND

The electroencephalogram (EEG) may be useful for monitoring anesthetic depth and avoiding overdose. We aimed to characterize EEG-recorded brain oscillations during increasing depth of anesthesia in a real-life surgical scenario. We hypothesized that alpha power and coherency will diminish as propofol dose increases between loss of consciousness (LOC) and an EEG burst suppression (BS) pattern.

METHODS

This nonrandomized dose-response clinical trial with concurrent control included EEG monitoring in 16 patients receiving slowly increasing doses of propofol. We assessed 3 intraoperative EEG segments (LOC, middle-dose, and BS) with spectral analysis.

RESULTS

Alpha band power diminished with each step increase in propofol dose. Average alpha power and average delta power during the BS step (-1.4±3.8 and 6.2±3.1 dB, respectively) were significantly lower than during the LOC step (2.8±2.6; P=0.004 and 10.1±5.2 dB; P=0.03, respectively). Peak alpha power was significantly higher during the LOC (5.4±2.6 dB) compared with middle-dose (2.6±3.6; P=0.04) and BS (0.7±3.2; P=0.0002) steps. In addition, as propofol dose increased, alpha band coherence between the F7 and F8 electrodes decreased, whereas delta band coherence exhibited a biphasic response (initial increase between LOC and middle-dose steps and decrease between middle-dose and BS steps).

CONCLUSION

We report compelling data regarding EEG patterns associated with increases in propofol dose. This information may more accurately define "therapeutic windows" for anesthesia and provide insights into brain dynamics that are sequentially affected by increased anesthetic doses.

Generalized Prediction of Unconsciousness during Propofol Anesthesia using 3D Convolutional Neural Networks

Neuroscience has generated a number of recent advances in the search for the neural correlates of consciousness, but these have yet to find valuable real-world applications. Electroencephalography under anesthesia provides a powerful experimental setup to identify electrophysiological signatures of altered states of consciousness, as well as a testbed for developing systems for automatic diagnosis and prognosis of awareness in clinical settings. In this work, we use deep convolutional neural networks to automatically differentiate sub-anesthetic states and depths of anesthesia, solely from one second of raw EEG signal. Our results with leave-one-participant-out-cross-validation show that behavioral measures, such as the Ramsay score, can be used to learn generalizable neural networks that reliably predict levels of unconsciousness in unseen transitional anesthetic states, as well as in unseen experimental setups and behaviors. Our findings highlight the potential of deep learning to detect progressive changes in anesthetic-induced unconsciousness with higher granularity than behavioral or pharmacological markers. This work has broader significance for identifying generalized patterns of brain activity that index states of consciousness.Clinical Relevance- In the United States alone, over 100,000 people receive general anesthesia every day, from which up to 1% is affected by unintended intraoperative awareness [1]. Despite this, brain-based monitoring of consciousness is not common in the clinic, and has had mixed success [2]. Given this context, our aim is to develop and explore an automated deep learning model that accurately predicts and interprets the depth and quality of anesthesia from the raw EEG signal.

Effect of Patient State Index Monitoring on the Recovery Characteristics in Morbidly Obese Patients: Comparison of Inhalation Anesthesia and Total Intravenous Anesthesia

PURPOSE

Obese patients have a significantly higher risk of adverse effects associated with general anesthesia. The purpose of this study was to evaluate the effects of Patient State Index (PSI) monitoring on recovery from anesthesia and the incidence of any postoperative complications among patients undergoing bariatric surgery with total intravenous anesthesia (TIVA) and inhalational anesthesia.

DESIGN

This prospective, double-blind, and randomized controlled trial was conducted between February 2017 and August 2017 and included 120 morbidly obese patients (body mass index >40 kg/m²).

METHODS

Patients were randomly divided into four groups; group P-PSI (n = 30): TIVA with PSI monitoring; group P (n = 30): TIVA without PSI monitoring; group D-PSI (n = 30): desflurane with PSI monitoring; and group D (n = 30): desflurane without PSI monitoring. The discharge time from the postanesthesia care unit (PACU), postoperative complications, and hemodynamic parameters were recorded and evaluated.

FINDINGS

No significant differences were found in demographic data, duration of anesthesia, admittance to PACU, discharge from PACU, modified Aldrete scores, and perioperative mean blood pressure and heart rate. Nausea and vomiting scores were significantly lower in group P-PSI, group P, and group D-PSI compared with group D.

CONCLUSIONS

Although TIVA and inhalational anesthesia can be safely used for obese patients, intraoperative PSI monitoring may decrease the discharge time from PACU and reduces incidence of postoperative nausea and vomiting caused by inhalation anesthetics.

Improved tracking of sevoflurane anesthetic states with drug-specific machine learning models

OBJECTIVE

The ability to monitor anesthetic states using automated approaches is expected to reduce inaccurate drug dosing and side-effects. Commercially available anesthetic state monitors perform poorly when ketamine is administered as an anesthetic-analgesic adjunct. Poor performance is likely because the models underlying these monitors are not optimized for the electroencephalogram (EEG) oscillations that are unique to the co-administration of ketamine.

APPROACH

In this work, we designed two k-nearest neighbors algorithms for anesthetic state prediction.

MAIN RESULTS

The first algorithm was trained only on sevoflurane EEG data, making it sevoflurane-specific. This algorithm enabled discrimination of the sevoflurane general anesthesia (GA) state from sedated and awake states (true positive rate = 0.87, [95% CI, 0.76, 0.97]). However, it did not enable discrimination of the sevoflurane-plus-ketamine GA state from sedated and awake states (true positive rate = 0.43, [0.19, 0.67]). In our second algorithm, we implemented a cross drug training paradigm by including both sevoflurane and sevoflurane-plus-ketamine EEG data in our training set. This algorithm enabled discrimination of the sevoflurane-plus-ketamine GA state from sedated and awake states (true positive rate = 0.91, [0.84, 0.98]).

SIGNIFICANCE

Instead of a one-algorithm-fits-all-drugs approach to anesthetic state monitoring, our results suggest that drug-specific models are necessary to improve the performance of automated anesthetic state monitors.

EEG-Brain Activity Monitoring and Predictive Analysis of Signals Using Artificial Neural Networks

Predictive observation and real-time analysis of the values of biomedical signals and automatic detection of epileptic seizures before onset are beneficial for the development of warning systems for patients because the patient, once informed that an epilepsy seizure is about to start, can take safety measures in useful time. In this article, Daubechies discrete wavelet transform (DWT) was used, coupled with analysis of the correlations between biomedical signals that measure the electrical activity in the brain by electroencephalogram (EEG), electrical currents generated in muscles by electromyogram (EMG), and heart rate monitoring by photoplethysmography (PPG). In addition, we used artificial neural networks (ANN) for automatic detection of epileptic seizures before onset. We analyzed 30 EEG recordings 10 min before a seizure and during the seizure for 30 patients with epilepsy. In this work, we investigated the ANN dimensions of 10, 50, 100, and 150 neurons, and we found that using an ANN with 150 neurons generates an excellent performance in comparison to a 10-neuron-based ANN. However, this analyzes requests in an increased amount of time in comparison with an ANN with a lower neuron number. For real-time monitoring, the neurons number should be correlated with the response time and power consumption used in wearable devices.

Spectral and Entropic Features Are Altered by Age in the Electroencephalogram in Patients under Sevoflurane Anesthesia

BACKGROUND

Preexisting factors such as age and cognitive performance can influence the electroencephalogram (EEG) during general anesthesia. Specifically, spectral EEG power is lower in elderly, compared to younger, subjects. Here, the authors investigate age-related changes in EEG architecture in patients undergoing general anesthesia through a detailed examination of spectral and entropic measures.

METHODS

The authors retrospectively studied 180 frontal EEG recordings from patients undergoing general anesthesia, induced with propofol/fentanyl and maintained by sevoflurane at the Waikato Hospital in Hamilton, New Zealand. The authors calculated power spectral density and normalized power spectral density, the entropic measures approximate and permutation entropy, as well as the beta ratio and spectral entropy as exemplary parameters used in current monitoring systems from segments of EEG obtained before the onset of surgery (i.e., with no noxious stimulation).

RESULTS

The oldest quartile of patients had significantly lower 1/f characteristics (P < 0.001; area under the receiver operating characteristics curve, 0.84 [0.76 0.92]), indicative of a more uniform distribution of spectral power. Analysis of the normalized power spectral density revealed no significant impact of age on relative alpha (P = 0.693; area under the receiver operating characteristics curve, 0.52 [0.41 0.63]) and a significant but weak effect on relative beta power (P = 0.041; area under the receiver operating characteristics curve, 0.62 [0.52 0.73]). Using entropic parameters, the authors found a significant age-related change toward a more irregular and unpredictable EEG (permutation entropy: P < 0.001, area under the receiver operating characteristics curve, 0.81 [0.71 0.90]; approximate entropy: P < 0.001; area under the receiver operating characteristics curve, 0.76 [0.66 0.85]). With approximate entropy, the authors could also detect an age-induced change in alpha-band activity (P = 0.002; area under the receiver operating characteristics curve, 0.69 [0.60 78]).

CONCLUSIONS

Like the sleep literature, spectral and entropic EEG features under general anesthesia change with age revealing a shift toward a faster, more irregular, oscillatory composition of the EEG in older patients. Age-related changes in neurophysiological activity may underlie these findings however the contribution of age-related changes in filtering properties or the signal to noise ratio must also be considered. Regardless, most current EEG technology used to guide anesthetic management focus on spectral features, and improvements to these devices might involve integration of entropic features of the raw EEG.

Photoplethysmography-derived approximate entropy and sample entropy as measures of analgesia depth during propofol-remifentanil anesthesia

The ability to monitor the physiological effect of the analgesic agent is of interest in clinical practice. Nonstationary changes would appear in photoplethysmography (PPG) during the analgesics-driven transition to analgesia. The present work studied the properties of nonlinear methods including approximate entropy (ApEn) and sample entropy (SampEn) derived from PPG responding to a nociceptive stimulus under various opioid concentrations. Forty patients with ASA I or II were randomized to receive one of the four possible remifentanil effect-compartment target concentrations (Ce_remi) of 0, 1, 3, and 5 ng·ml^-1 and a propofol effect-compartment target-controlled infusion to maintain the state entropy (SE) at 50 ± 10. Laryngeal mask airway (LMA) insertion was applied as a standard noxious stimulation. To optimize the performance of ApEn and SampEn, different coefficients were carefully evaluated. The monotonicity of ApEn and SampEn changing from low Ce_remi to high Ce_remi was assessed with prediction probabilities (P_K). The result showed that low Ce_remi (0 and 1 ng·ml^-1) could be differentiated from high Ce_remi (3 and 5 ng·ml^-1) by ApEn and SampEn. Depending on the coefficient employed in algorithm: ApEn with k = 0.15 yielded the largest P_K value (0.875) whereas SampEn gained its largest P_K of 0.867 with k = 0.2. Thus, PPG-based ApEn and SampEn with appropriate k values have the potential to offer good quantification of analgesia depth under general anesthesia.

Alpha rhythm collapse predicts iso-electric suppressions during anesthesia

Could an overly deep sedation be anticipated from ElectroEncephaloGram (EEG) patterns? We report here motifs hidden in the EEG signal that predict the appearance of Iso-Electric Suppressions (IES), observed during epileptic encephalopathies, drug intoxications, comatose, brain death or during anesthetic over-dosage that are considered to be detrimental. To show that IES occurrences can be predicted from EEG traces dynamics, we focus on transient suppression of the alpha rhythm (8-14 Hz) recorded for 80 patients, that had a Propofol target controlled infusion of 5 μg/ml during a general anesthesia. We found that the first time of appearance as well as changes in duration of these Alpha-Suppressions (αS) are two parameters that anticipate the appearance of IES. Using machine learning, we predicted IES appearance from the first 10 min of EEG (AUC of 0.93). To conclude, transient motifs in the alpha rhythm predict IES during anesthesia and can be used to identify patients, with higher risks of post-operative complications.

Design and Implementation of a Machine Learning Based EEG Processor for Accurate Estimation of Depth of Anesthesia

Accurate monitoring of the depth of anesthesia (DoA) is essential for intraoperative and postoperative patient's health. Commercially available electroencephalograph (EEG)-based DoA monitors are recommended only for certain anesthetic drugs and specific age-group patients. This paper presents a machine learning classification processor for accurate DoA estimation irrespective of the patient's age and anesthetic drug. The classification is solely based on six features extracted from EEG signal, i.e., spectral edge frequency (SEF), beta ratio, and four bands of spectral energy (FBSE). A machine learning fine decision tree classifier is adopted to achieve a four-class DoA classification (deep, moderate, and light DoA versus awake state). The feature selection and the classification processor are optimized to achieve the highest classification accuracy for the state of moderate anesthesia required for the surgical operations. The proposed 256-point fast Fourier transform accelerator is implemented to realize SEF, beta ratio, and FBSE that enables minimal latency and high accuracy feature extraction. The proposed DoA processor is implemented using a 65 nm CMOS technology and experimentally verified using field programming gate array (FPGA) based on the EEG recordings of 75 patients undergoing elective surgery with different types of anesthetic agents. The processor achieves an average accuracy of 92.2% for all DoA states, with a latency of 1s The 0.09 mm² DoA processor consumes 140nJ/classification.

Drug-Specific Models Improve the Performance of an EEG-based Automated Brain-State Prediction System

Maintaining anesthetic states using automated brain-state prediction systems is expected to reduce drug overdosage and associated side-effects. However, commercially available brain-state monitoring systems perform poorly on drug-class combinations. We assume that current automated brain-state prediction systems perform poorly because they do not account for brain-state dynamics that are unique to drug-class combinations. In this work, we develop a k-nearest neighbors model to test whether improvements to automated brain-state prediction of drug-class combinations are feasible. We utilize electroencephalogram data collected from human subjects who received general anesthesia with sevoflurane and general anesthesia with the drug-class combination of sevoflurane-plus-ketamine. We demonstrate improved performance predicting anesthesia-induced brain-states using drug-specific models.

EEG power spectral density under Propofol and its association with burst suppression, a marker of cerebral fragility

OBJECTIVE

Under General Anesthesia (GA), age and Burst Suppression (BS) are associated with cognitive postoperative complications, yet how these parameters are related to per-operative EEG and hypnotic doses is unclear. In this prospective study, we address this question comparing age and BS occurrences with a new score (BP_TIVA) based on Propofol doses, EEG and alpha-band power spectral densities, evaluated for SEF₉₅ = 8-13 Hz.

METHODS

59 patients (55 [34-67] yr, 67% female) undergoing neuroradiology or orthopedic surgery were included. Total IntraVenous Anesthesia was used for Propofol and analgesics infusion. Cerebral activity was monitored from a frontal electrodes montage EEG.

RESULTS

BP_TIVA was inversely correlated with age (Pearson r = -0.78, p < 0.001), and was significantly lower (p < 0.001) when BS occurred during the GA first minutes (induction). Additionally, the age-free BP_TIVA score was better associated with BS at induction than age (AUC = 0.94 versus 0.82, p < 0.05).

CONCLUSION

We designed BP_TIVA score based on hypnotics and EEG. It was correlated with age yet was better associated to BS occurring during GA induction, the latter being a cerebral fragility sign.

SIGNIFICANCE

This advocate for an approach based on evaluating the cerebral physiological age («brain age») to predict postoperative cognitive evolution.

Neural Correlates of Anesthesia in Newborn Mice and Humans

Monitoring the hypnotic component of anesthesia during surgeries is critical to prevent intraoperative awareness and reduce adverse side effects. For this purpose, electroencephalographic (EEG) methods complementing measures of autonomic functions and behavioral responses are in use in clinical practice. However, in human neonates and infants existing methods may be unreliable and the correlation between brain activity and anesthetic depth is still poorly understood. Here, we characterized the effects of different anesthetics on brain activity in neonatal mice and developed machine learning approaches to identify electrophysiological features predicting inspired or end-tidal anesthetic concentration as a proxy for anesthetic depth. We show that similar features from EEG recordings can be applied to predict anesthetic concentration in neonatal mice and humans. These results might support a novel strategy to monitor anesthetic depth in human newborns.

Calculations of consciousness: electroencephalography analyses to determine anesthetic depth

PURPOSE OF REVIEW

Electroencephalography (EEG) was introduced into anesthesia practice in the 1990s as a tool to titrate anesthetic depth. However, limitations in current analysis techniques have called into question whether these techniques improve standard of care, or instead call for improved, more ubiquitously applicable measures to assess anesthetic transitions and depth. This review highlights emerging analytical approaches and techniques from neuroscience research that have the potential to better capture anesthetic transitions to provide better measurements of anesthetic depth.

RECENT FINDINGS

Since the introduction of electroencephalography, neuroscientists, engineers, mathematicians, and clinicians have all been developing new ways of analyzing continuous electrical signals. Collaborations between these fields have proliferated several analytical techniques that demonstrate how anesthetics affect brain dynamics and conscious transitions. Here, we review techniques in the following categories: network science, integration and information, nonlinear dynamics, and artificial intelligence.

SUMMARY

Up-and-coming techniques have the potential to better clinically define and characterize altered consciousness time points. Such new techniques used alongside traditional measures have the potential to improve depth of anesthesia measurements and enhance an understanding of how the brain is affected by anesthetic agents. However, new measures will be needed to be tested for robustness in real-world environments and on diverse experimental protocols.

Automatic anesthesia depth staging using entropy measures and relative power of electroencephalogram frequency bands

Many of the surgeries performed under general anesthesia are aided by electroencephalogram (EEG) monitoring. With increased focus on detecting the anesthesia states of patients in the course of surgery, more attention has been paid to analyzing the power spectra and entropy measures of EEG signal during anesthesia. In this paper, by using the relative power of EEG frequency bands and the EEG entropy measures, a new method for detecting the depth of anesthesia states has been presented based on the least squares support vector machine (LS-SVM) classifiers. EEG signals were recorded from 20 patients before, during and after general anesthesia in the operating room at a sampling rate of 200 Hz. Then, 12 features were extracted from each EEG segment, 10 s in length, which are used for anesthesia state monitoring. No significant difference was observed (p > 0.05) between these features and the bispectral index (BIS), which is the commonly used measure of anesthetic effect. The used LS-SVM classifier based method is able to identify the anesthesia states with an accuracy of 80% with reference to the BIS index. Since the underlying equation of the utilized LS-SVM is linear, the computational time of the algorithm is not significant and therefore it can be used for online application in operation rooms.

Distinguishing Anesthetized from Awake State in Patients: A New Approach Using One Second Segments of Raw EEG

Objective: The objective of this study was to test whether properties of 1-s segments of spontaneous scalp EEG activity can be used to automatically distinguish the awake state from the anesthetized state in patients undergoing general propofol anesthesia.

Methods: Twenty five channel EEG was recorded from 10 patients undergoing general intravenous propofol anesthesia with remifentanil during anterior cervical discectomy and fusion. From this, we extracted properties of the EEG by applying the Directed Transfer Function (DTF) directly to every 1-s segment of the raw EEG signal. The extracted properties were used to develop a data-driven classification algorithm to categorize patients as “anesthetized” or “awake” for every 1-s segment of raw EEG.

Results: The properties of the EEG signal were significantly different in the awake and anesthetized states for at least 8 of the 25 channels (p < 0.05, Bonferroni corrected Wilcoxon rank-sum tests). Using these differences, our algorithms achieved classification accuracies of 95.9%.

Conclusion: Properties of the DTF calculated from 1-s segments of raw EEG can be used to reliably classify whether the patients undergoing general anesthesia with propofol and remifentanil were awake or anesthetized.

Significance: This method may be useful for developing automatic real-time monitors of anesthesia.