Electroencephalogram-Based Complexity Measures as Predictors of Post-operative Neurocognitive Dysfunction

Identifying EEG Biomarkers of Depression with Novel Explainable Deep Learning Architectures

Deep learning methods are increasingly being applied to raw electroencephalogram (EEG) data. However, if these models are to be used in clinical or research contexts, methods to explain them must be developed, and if these models are to be used in research contexts, methods for combining explanations across large numbers of models must be developed to counteract the inherent randomness of existing training approaches. Model visualization-based explainability methods for EEG involve structuring a model architecture such that its extracted features can be characterized and have the potential to offer highly useful insights into the patterns that they uncover. Nevertheless, model visualization-based explainability methods have been underexplored within the context of multichannel EEG, and methods to combine their explanations across folds have not yet been developed. In this study, we present two novel convolutional neural network-based architectures and apply them for automated major depressive disorder diagnosis. Our models obtain slightly lower classification performance than a baseline architecture. However, across 50 training folds, they find that individuals with MDD exhibit higher β power, potentially higher δ power, and higher brain-wide correlation that is most strongly represented within the right hemisphere. This study provides multiple key insights into MDD and represents a significant step forward for the domain of explainable deep learning applied to raw EEG. We hope that it will inspire future efforts that will eventually enable the development of explainable EEG deep learning models that can contribute both to clinical care and novel medical research discoveries.

Intraoperative monitoring of the central and peripheral nervous systems: a narrative review

The central and peripheral nervous systems are the primary target organs during anaesthesia. At the time of the inception of the British Journal of Anaesthesia, monitoring of the central nervous system comprised clinical observation, which provided only limited information. During the 100 yr since then, and particularly in the past few decades, significant progress has been made, providing anaesthetists with tools to obtain real-time assessments of cerebral neurophysiology during surgical procedures. In this narrative review article, we discuss the rationale and uses of electroencephalography, evoked potentials, near-infrared spectroscopy, and transcranial Doppler ultrasonography for intraoperative monitoring of the central and peripheral nervous systems.

Update of the European Society of Anaesthesiology and Intensive Care Medicine evidence-based and consensus-based guideline on postoperative delirium in adult patients

Postoperative delirium (POD) remains a common, dangerous and resource-consuming adverse event but is often preventable. The whole peri-operative team can play a key role in its management. This update to the 2017 ESAIC Guideline on the prevention of POD is evidence-based and consensus-based and considers the literature between 01 April 2015, and 28 February 2022. The search terms of the broad literature search were identical to those used in the first version of the guideline published in 2017. POD was defined in accordance with the DSM-5 criteria. POD had to be measured with a validated POD screening tool, at least once per day for at least 3 days starting in the recovery room or postanaesthesia care unit on the day of surgery or, at latest, on postoperative day 1. Recent literature confirmed the pathogenic role of surgery-induced inflammation, and this concept reinforces the positive role of multicomponent strategies aimed to reduce the surgical stress response. Although some putative precipitating risk factors are not modifiable (length of surgery, surgical site), others (such as depth of anaesthesia, appropriate analgesia and haemodynamic stability) are under the control of the anaesthesiologists. Multicomponent preoperative, intra-operative and postoperative preventive measures showed potential to reduce the incidence and duration of POD, confirming the pivotal role of a comprehensive and team-based approach to improve patients' clinical and functional status.

Preoperative electroencephalographic alpha-power changes with eyes opening are associated with postoperative attention impairment and inattention-related delirium severity

BACKGROUND

In the eyes-closed, awake condition, EEG oscillatory power in the alpha band (7-13 Hz) dominates human spectral activity. With eyes open, however, EEG alpha power substantially decreases. Less alpha attenuation with eyes opening has been associated with inattention; thus, we analysed whether reduced preoperative alpha attenuation with eyes opening is associated with postoperative inattention, a delirium-defining feature.

METHODS

Preoperative awake 32-channel EEG was recorded with eyes open and eyes closed in 71 non-neurological, noncardiac surgery patients aged ≥ 60 years. Inattention and other delirium features were assessed before surgery and twice daily after surgery until discharge. Eyes-opening EEG alpha-attenuation magnitude was analysed for associations with postoperative inattention, primarily, and with delirium severity, secondarily, using multivariate age- and Mini-Mental Status Examination (MMSE)-adjusted logistic and proportional-odds regression analyses.

RESULTS

Preoperative alpha attenuation with eyes opening was inversely associated with postoperative inattention (odds ratio [OR] 0.73, 95% confidence interval [CI]: 0.57, 0.94; P=0.038). Sensitivity analyses showed an inverse relationship between alpha-attenuation magnitude and inattention chronicity, defined as 'never', 'newly', or 'chronically' inattentive (OR 0.76, 95% CI: 0.62, 0.93; P=0.019). In addition, preoperative alpha-attenuation magnitude was inversely associated with postoperative delirium severity (OR 0.79, 95% CI: 0.65, 0.95; P=0.040), predominantly as a result of the inattention feature.

CONCLUSIONS

Preoperative awake, resting, EEG alpha attenuation with eyes opening might represent a neural biomarker for risk of postoperative attentional impairment. Further, eyes-opening alpha attenuation could provide insight into the neural mechanisms underlying postoperative inattention risk.

Electroencephalographic Biomarkers, Cerebral Oximetry, and Postoperative Cognitive Function in Adult Noncardiac Surgical Patients: A Prospective Cohort Study

BACKGROUND

Perioperative neurocognitive disorders are a major public health issue, although there are no validated neurophysiologic biomarkers that predict cognitive function after surgery. This study tested the hypothesis that preoperative posterior electroencephalographic alpha power, alpha frontal-parietal connectivity, and cerebral oximetry would each correlate with postoperative neurocognitive function.

METHODS

This was a single-center, prospective, observational study of adult (older than 18 yr) male and female noncardiac surgery patients. Whole-scalp, 16-channel electroencephalography and cerebral oximetry were recorded in the preoperative, intraoperative, and immediate postoperative settings. The primary outcome was the mean postoperative T-score of three National Institutes of Health Toolbox Cognition tests-Flanker Inhibitory Control and Attention, List Sorting Working Memory, and Pattern Comparison Processing Speed. These tests were obtained at preoperative baseline and on the first two postoperative mornings. The lowest average score from the first two postoperative days was used for the primary analysis. Delirium was a secondary outcome (via 3-min Confusion Assessment Method) measured in the postanesthesia care unit and twice daily for the first 3 postoperative days. Last, patient-reported outcomes related to cognition and overall well-being were collected 3 months postdischarge.

RESULTS

Sixty-four participants were recruited with a median (interquartile range) age of 59 (48 to 66) yr. After adjustment for baseline cognitive function scores, no significant partial correlation (ρ) was detected between postoperative cognition scores and preoperative relative posterior alpha power (%; ρ = -0.03, P = 0.854), alpha frontal-parietal connectivity (via weight phase lag index; ρ = -0.10, P = 0.570, respectively), or preoperative cerebral oximetry (%; ρ = 0.21, P = 0.246). Only intraoperative frontal-parietal theta connectivity was associated with postoperative delirium (F[1,6,291] = 4.53, P = 0.034). No electroencephalographic or oximetry biomarkers were associated with cognitive or functional outcomes 3 months postdischarge.

CONCLUSIONS

Preoperative posterior alpha power, frontal-parietal connectivity, and cerebral oximetry were not associated with cognitive function after noncardiac surgery.

EDITOR’S PERSPECTIVE

EEG microstate quantifiers and state space descriptors during anaesthesia in patients with postoperative delirium: a descriptive analysis

Postoperative delirium is a serious sequela of surgery and surgery-related anaesthesia. One recommended method to prevent postoperative delirium is using bi-frontal EEG recording. The single, processed index of depth of anaesthesia allows the anaesthetist to avoid episodes of suppression EEG and excessively deep anaesthesia. The study data presented here were based on multichannel (19 channels) EEG recordings during anaesthesia. This enabled the analysis of various parameters of global electrical brain activity. These parameters were used to compare microstate topographies under anaesthesia with those in healthy volunteers and to analyse changes in microstate quantifiers and EEG global state space descriptors with increasing exposure to anaesthesia. Seventy-three patients from the Surgery Depth of Anaesthesia and Cognitive Outcome study (SRCTN 36437985) received intraoperative multichannel EEG recordings. Altogether, 720 min of artefact-free EEG data, including 210 min (29.2%) of suppression EEG, were analysed. EEG microstate topographies, microstate quantifiers (duration, frequency of occurrence and global field power) and the state space descriptors sigma (overall EEG power), phi (generalized frequency) and omega (number of uncorrelated brain processes) were evaluated as a function of duration of exposure to anaesthesia, suppression EEG and subsequent development of postoperative delirium. The major analyses involved covariate-adjusted linear mixed-effects models. The older (71 ± 7 years), predominantly male (60%) patients received a median exposure of 210 (range: 75-675) min of anaesthesia. During seven postoperative days, 21 patients (29%) developed postoperative delirium. Microstate topographies under anaesthesia resembled topographies from healthy and much younger awake persons. With increasing duration of exposure to anaesthesia, single microstate quantifiers progressed differently in suppression or non-suppression EEG and in patients with or without subsequent postoperative delirium. The most pronounced changes occurred during enduring suppression EEG in patients with subsequent postoperative delirium: duration and frequency of occurrence of microstates C and D progressed in opposite directions, and the state space descriptors showed a pattern of declining uncorrelated brain processes (omega) combined with increasing EEG variance (sigma). With increasing exposure to general anaesthesia, multiple changes in the dynamics of microstates and global EEG parameters occurred. These changes varied partly between suppression and non-suppression EEG and between patients with or without subsequent postoperative delirium. Ongoing suppression EEG in patients with subsequent postoperative delirium was associated with reduced network complexity in combination with increased overall EEG power. Additionally, marked changes in quantifiers in microstate C and in microstate D occurred. These putatively adverse intraoperative trajectories in global electrical brain activity may be seen as preceding and ultimately predicting postoperative delirium.

EEG pre-burst suppression: characterization and inverse association with preoperative cognitive function in older adults

The most common complication in older surgical patients is postoperative delirium (POD). POD is associated with preoperative cognitive impairment and longer durations of intraoperative burst suppression (BSup) - electroencephalography (EEG) with repeated periods of suppression (very low-voltage brain activity). However, BSup has modest sensitivity for predicting POD. We hypothesized that a brain state of lowered EEG power immediately precedes BSup, which we have termed "pre-burst suppression" (preBSup). Further, we hypothesized that even patients without BSup experience these preBSup transient reductions in EEG power, and that preBSup (like BSup) would be associated with preoperative cognitive function and delirium risk. Data included 83 32-channel intraoperative EEG recordings of the first hour of surgery from 2 prospective cohort studies of patients ≥age 60 scheduled for ≥2-h non-cardiac, non-neurologic surgery under general anesthesia (maintained with a potent inhaled anesthetic or a propofol infusion). Among patients with BSup, we defined preBSup as the difference in 3-35 Hz power (dB) during the 1-s preceding BSup relative to the average 3-35 Hz power of their intraoperative EEG recording. We then recorded the percentage of time that each patient spent in preBSup, including those without BSup. Next, we characterized the association between percentage of time in preBSup and (1) percentage of time in BSup, (2) preoperative cognitive function, and (3) POD incidence. The percentage of time in preBSup and BSup were correlated (Spearman's ρ [95% CI]: 0.52 [0.34, 0.66], p < 0.001). The percentage of time in BSup, preBSup, or their combination were each inversely associated with preoperative cognitive function (β [95% CI]: -0.10 [-0.19, -0.01], p = 0.024; -0.04 [-0.06, -0.01], p = 0.009; -0.04 [-0.06, -0.01], p = 0.003, respectively). Consistent with prior literature, BSup was significantly associated with POD (odds ratio [95% CI]: 1.34 [1.01, 1.78], p = 0.043), though this association did not hold for preBSup (odds ratio [95% CI]: 1.04 [0.95, 1.14], p = 0.421). While all patients had ≥1 preBSup instance, only 20.5% of patients had ≥1 BSup instance. These exploratory findings suggest that future studies are warranted to further study the extent to which preBSup, even in the absence of BSup, can identify patients with impaired preoperative cognition and/or POD risk.

A Real-Time Neurophysiologic Stress Test for the Aging Brain: Novel Perioperative and ICU Applications of EEG in Older Surgical Patients

As of 2022, individuals age 65 and older represent approximately 10% of the global population [1], and older adults make up more than one third of anesthesia and surgical cases in developed countries [2, 3]. With approximately > 234 million major surgical procedures performed annually worldwide [4], this suggests that > 70 million surgeries are performed on older adults across the globe each year. The most common postoperative complications seen in these older surgical patients are perioperative neurocognitive disorders including postoperative delirium, which are associated with an increased risk for mortality [5], greater economic burden [6, 7], and greater risk for developing long-term cognitive decline [8] such as Alzheimer's disease and/or related dementias (ADRD). Thus, anesthesia, surgery, and postoperative hospitalization have been viewed as a biological "stress test" for the aging brain, in which postoperative delirium indicates a failed stress test and consequent risk for later cognitive decline (see Fig. 3). Further, it has been hypothesized that interventions that prevent postoperative delirium might reduce the risk of long-term cognitive decline. Recent advances suggest that rather than waiting for the development of postoperative delirium to indicate whether a patient "passed" or "failed" this stress test, the status of the brain can be monitored in real-time via electroencephalography (EEG) in the perioperative period. Beyond the traditional intraoperative use of EEG monitoring for anesthetic titration, perioperative EEG may be a viable tool for identifying waveforms indicative of reduced brain integrity and potential risk for postoperative delirium and long-term cognitive decline. In principle, research incorporating routine perioperative EEG monitoring may provide insight into neuronal patterns of dysfunction associated with risk of postoperative delirium, long-term cognitive decline, or even specific types of aging-related neurodegenerative disease pathology. This research would accelerate our understanding of which waveforms or neuronal patterns necessitate diagnostic workup and intervention in the perioperative period, which could potentially reduce postoperative delirium and/or dementia risk. Thus, here we present recommendations for the use of perioperative EEG as a "predictor" of delirium and perioperative cognitive decline in older surgical patients.

Preclinical and translational models for delirium: Recommendations for future research from the NIDUS delirium network

Delirium is a common, morbid, and costly syndrome that is closely linked to Alzheimer's disease (AD) and AD-related dementias (ADRD) as a risk factor and outcome. Human studies of delirium have advanced our knowledge of delirium incidence and prevalence, risk factors, biomarkers, outcomes, prevention, and management. However, understanding of delirium neurobiology remains limited. Preclinical and translational models for delirium, while challenging to develop, could advance our knowledge of delirium neurobiology and inform the development of new prevention and treatment approaches. We discuss the use of preclinical and translational animal models in delirium, focusing on (1) a review of current animal models, (2) challenges and strategies for replicating elements of human delirium in animals, and (3) the utility of biofluid, neurophysiology, and neuroimaging translational markers in animals. We conclude with recommendations for the development and validation of preclinical and translational models for delirium, with the goal of advancing awareness in this important field.

NOVEL APPROACH EXPLAINS SPATIO-SPECTRAL INTERACTIONS IN RAW ELECTROENCEPHALOGRAM DEEP LEARNING CLASSIFIERS

The application of deep learning classifiers to resting-state electroencephalography (rs-EEG) data has become increasingly common. However, relative to studies using traditional machine learning methods and extracted features, deep learning methods are less explainable. A growing number of studies have presented explainability approaches for rs-EEG deep learning classifiers. However, to our knowledge, no approaches give insight into spatio-spectral interactions (i.e., how spectral activity in one channel may interact with activity in other channels). In this study, we combine gradient and perturbation-based explainability approaches to give insight into spatio-spectral interactions in rs-EEG deep learning classifiers for the first time. We present the approach within the context of major depressive disorder (MDD) diagnosis identifying differences in frontal δ activity and reduced interactions between frontal electrodes and other electrodes. Our approach provides novel insights and represents a significant step forward for the field of explainable EEG classification.

Integration of multiscale entropy and BASED scale of electroencephalography after adrenocorticotropic hormone therapy predict relapse of infantile spasms

BACKGROUND

Even though adrenocorticotropic hormone (ACTH) demonstrated powerful efficacy in the initially successful treatment of infantile spasms (IS), nearly half of patients have experienced a relapse. We sought to investigate whether features of electroencephalogram (EEG) predict relapse in those IS patients without structural brain abnormalities.

METHODS

We retrospectively reviewed data from children with IS who achieved initial response after ACTH treatment, along with EEG recorded within the last two days of treatment. The recurrence of epileptic spasms following treatment was tracked for 12 months. Subjects were categorized as either non-relapse or relapse groups. General clinical and EEG recordings were collected, burden of amplitudes and epileptiform discharges (BASED) score and multiscale entropy (MSE) were carefully explored for cross-group comparisons.

RESULTS

Forty-one patients were enrolled in the study, of which 26 (63.4%) experienced a relapse. The BASED score was significantly higher in the relapse group. MSE in the non-relapse group was significantly lower than the relapse group in the γ band but higher in the lower frequency range (δ, θ, α). Sensitivity and specificity were 85.71% and 92.31%, respectively, when combining MSE in the δ/γ frequency of the occipital region, plus BASED score were used to distinguish relapse from non-relapse groups.

CONCLUSIONS

BASED score and MSE of EEG after ACTH treatment could be used to predict relapse for IS patients without brain structural abnormalities. Patients with BASED score ≥ 3, MSE increased in higher frequency, and decreased in lower frequency had a high risk of relapse.

Electroencephalogram Features of Perioperative Neurocognitive Disorders in Elderly Patients: A Narrative Review of the Clinical Literature

Postoperative neurocognitive disorder (PND) is a common postoperative complication, particularly in older patients. Electroencephalogram (EEG) monitoring, a non-invasive technique with a high spatial-temporal resolution, can accurately characterize the dynamic changes in brain function during the perioperative period. Current clinical studies have confirmed that the power density of alpha oscillation during general anesthesia decreased with age, which was considered to be associated with increased susceptibility to PND in the elderly. However, evidence on whether general anesthesia under EEG guidance results in a lower morbidity of PND is still contradictory. This is one of the reasons that common indicators of the depth of anesthesia were limitedly derived from EEG signals in the frontal lobe. The variation of multi-channel EEG features during the perioperative period has the potential to highlight the occult structural and functional abnormalities of the subcortical-cortical neurocircuit. Therefore, we present a review of the application of multi-channel EEG monitoring to predict the incidence of PND in older patients. The data confirmed that the abnormal variation in EEG power and functional connectivity between distant brain regions was closely related to the incidence and long-term poor outcomes of PND in older adults.

Multiscale entropy analysis of retinal signals reveals reduced complexity in a mouse model of Alzheimer's disease

Alzheimer's disease (AD) is one of the most significant health challenges of our time, affecting a growing number of the elderly population. In recent years, the retina has received increased attention as a candidate for AD biomarkers since it appears to manifest the pathological signatures of the disease. Therefore, its electrical activity may hint at AD-related physiological changes. However, it is unclear how AD affects retinal electrophysiology and what tools are more appropriate to detect these possible changes. In this study, we used entropy tools to estimate the complexity of the dynamics of healthy and diseased retinas at different ages. We recorded microelectroretinogram responses to visual stimuli of different nature from retinas of young and adult, wild-type and 5xFAD-an animal model of AD-mice. To estimate the complexity of signals, we used the multiscale entropy approach, which calculates the entropy at several time scales using a coarse graining procedure. We found that young retinas had more complex responses to different visual stimuli. Further, the responses of young, wild-type retinas to natural-like stimuli exhibited significantly higher complexity than young, 5xFAD retinas. Our findings support a theory of complexity-loss with aging and disease and can have significant implications for early AD diagnosis.