Barry J. Ruijter

Generalized epileptiform discharges in postanoxic encephalopathy: Quantitative characterization in relation to outcome

Electrographic status epilepticus is observed in 10–35% of patients with postanoxic encephalopathy. It remains unclear which electrographic seizure patterns indicate possible recovery, and which are a mere reflection of severe ischemic encephalopathy, where treatment would be futile. We aimed to identify quantitative electroencephalography (EEG) features with prognostic significance.

Synaptic damage underlies EEG abnormalities in postanoxic encephalopathy: A computational study

OBJECTIVE In postanoxic coma, EEG patterns indicate the severity of encephalopathy and typically evolve in time. We aim to improve the understanding of pathophysiological mechanisms underlying these EEG abnormalities. METHODS We used a mean field model comprising excitatory and inhibitory neurons, local synaptic connections, and input from thalamic afferents. Anoxic damage is modeled as aggravated short-term synaptic depression, with gradual recovery over many hours. Additionally, excitatory neurotransmission is potentiated, scaling with the severity of anoxic encephalopathy. Simulations were compared with continuous EEG recordings of 155 comatose patients after cardiac arrest. RESULTS The simulations agree well with six common categories of EEG rhythms in postanoxic encephalopathy, including typical transitions in time. Plausible results were only obtained if excitatory synapses were more severely affected by short-term synaptic depression than inhibitory synapses. CONCLUSIONS In postanoxic encephalopathy, the evolution of EEG patterns presumably results from gradual improvement of complete synaptic failure, where excitatory synapses are more severely affected than inhibitory synapses. The range of EEG patterns depends on the excitation-inhibition imbalance, probably resulting from long-term potentiation of excitatory neurotransmission. SIGNIFICANCE Our study is the first to relate microscopic synaptic dynamics in anoxic brain injury to both typical EEG observations and their evolution in time.

Early electroencephalography dynamics after cardiac arrest

To the Editor:With interest we read the article published in a recent issue of Critical Care Medicine by Rossetti et al (1), in which they confirm that early electroencephalography (EEG) contributes to outcome prediction of postanoxic coma. We concede the importance of external validation of previou

The Revised Cerebral Recovery Index Improves Predictions of Neurological Outcome after Cardiac Arrest

OBJECTIVE Analysis of the electroencephalogram (EEG) background pattern helps predicting neurological outcome of comatose patients after cardiac arrest (CA). Visual analysis may not extract all discriminative information. We present predictive values of the revised Cerebral Recovery Index (rCRI), based on continuous extraction and combination of a large set of evolving quantitative EEG (qEEG) features and machine learning techniques. METHODS We included 551 subsequent patients from a prospective cohort study on continuous EEG after CA in two hospitals. Outcome at six months was classified as good (Cerebral Performance Category (CPC) 1-2) or poor (CPC 3-5). Forty-four qEEG features (from time, frequency and entropy domain) were selected by the least absolute shrinkage and selection operator (LASSO) method and used in a Random Forests classification system. We trained and evaluated the system with 10-fold cross validation. For poor outcome prediction, the sensitivity at 100% specificity (Se100) and the area under the receiver operator curve (AUC) were used as performance of the prediction model. For good outcome, we used the sensitivity at 95% specificity (Se95). RESULTS Two hundred fifty-six (47%) patients had a good outcome. The rCRI predicted poor outcome with AUC = 0.94 (95% CI: 0.83-0.91), Se100 = 0.66 (0.65-0.78), and AUC = 0.88 (0.78-0.93), Se100 = 0.60 (0.51-0.75) at 12 and 24 h after CA, respectively. The rCRI predicted good outcome with Se95 = 0.72 (0.61-0.85) and 0.40 (0.30-0.51) at 12 and 24 h after CA, respectively. CONCLUSIONS Results obtained in this study suggest that with machine learning algorithms and large set of qEEG features, it is possible to efficiently monitor patient outcome after CA. We also demonstrate the importance of selection of optimal performance metric to train a classifier model for outcome prediction. SIGNIFICANCE The rCRI is a sensitive, reliable predictor of neurological outcome of comatose patients after CA.

Platform Session – Electroencephalography/Epilepsy: EEG for the prediction of outcome within the first five days of postanoxic coma: A prospective multicenter cohort study

Introduction We recently showed that the EEG at 12 and 24 h after cardiac arrest reliably predicts the outcome of postanoxic coma. In this study, we investigate whether outcome predictions could be further improved by adding unfavorable EEG categories, and by evaluation of the EEG at other time points in the first five days after cardiac arrest. We validate our findings in a larger cohort, with patients recruited from three additional centers. Methods Data were collected in a prospective cohort study conducted at intensive care units of five hospitals in the Netherlands, between 2010 and 2017. In all comatose survivors of cardiac arrest, continuous EEG recordings were started as soon as possible and continued up to five days. Five-minute EEG epochs at 6, 12, 24, 36, 48, 72, 96, and 120 h after cardiac arrest were assessed by two independent reviewers, blinded for patients’ outcome. EEG classification was based on the American Clinical Neurophysiology Society’s standardized critical care EEG terminology. Patterns were categorized as favorable (continuous or nearly continuous background, with normal voltage) or unfavorable (burst-suppression with identical bursts or generalized periodic discharges (GPDs) on a suppressed background). A suppressed EEG (voltage  μ V) at 12 h or later, and a low-voltage EEG (10–20  μ V) at 24 h or later, were also considered unfavorable. Outcome at 6 months after cardiac arrest was categorized as good (Cerebral Performance Category (CPC) 1–2) or poor (CPC 3–5). Results At time of submission of this abstract, data from two participating sites (Medisch Spectrum Twente, Rijnstate) have been analyzed. In these centers, 569 patients were included, of which 46% had a good outcome. The best timing to predict either a good or poor outcome was 12 h after cardiac arrest. A favorable EEG at 12 h predicted good outcome with a sensitivity of 0.46 (95% confidence interval (CI): 0.38–0.54) at a specificity of 0.90 (95%-CI: 0.85–0.94). At 24 h or later, specificity for the prediction of good outcome dropped below 0.90. At any of the investigated time points, an unfavorable EEG pattern predicted poor outcome without false positives. At 12 h, sensitivity reached its maximum value of 0.45 (95%-CI: 0.37–0.52) at specificity of 1.00 (95%-CI: 0.98–1.00) and dropped below 0.30 at 36 h or later. Conclusion We confirm that relevant discrimination for the prediction of good outcome after cardiac arrest with EEG is only possible up to 12 h after the event. Poor outcome can be predicted reliably up to five days, but sensitivity decreases significantly after the first 24 h. As compared to previous work, the addition of a suppressed EEG at 12 h and GPDs on a suppressed background at any time as unfavorable patterns increases sensitivity for the prediction of poor outcome. We will validate our findings on data from an additional 250 patients, recruited at the other three sites, and present these results at the ICCN 2018.

S27. Outcome prediction in postanoxic coma with deep learning

Introduction Visual assessment of the electroencephalogram (EEG) by experienced clinical neurophysiologists allows reliable outcome prediction in up to half of all comatose patients after cardiac arrest. We hypothesize that deep neural networks can achieve similar or better performance, while being objective and consistent. Methods In a prospective cohort study, continuous EEG recordings from comatose patients after cardiac arrest were collected from the intensive care units of two large teaching hospitals. Functional outcome at six months was assessed using the Cerebral Performance Category scale (CPC), dichotomized as good (CPC 1–2) or poor (CPC 3–5). Five-minute artifact-free EEG epochs at 12 and 24 h after cardiac arrest were partitioned into 10 s epochs. We trained a convolutional neural network, using the raw EEG epochs and outcome labels as inputs to predict outcome using data from 80% of the patients. Validation was performed in the remaining 20%. The probability of recovery to good neurological outcome was quantified for each individual patient. Analyses of diagnostic accuracy included receiver operating characteristics and calculation of predictive values at 12 and 24 h. Results Four hundred and fifty-six patients were included, resulting in 306 and 439 EEGs epochs at 12 and 24 h, respectively. Outcome prediction was most accurate at 12 h, with an area under the ROC curve (AUC) of 0.89 versus 0.81 at 24 h. Poor outcome could be predicted at 12 h with a sensitivity of 62% (95% confidence interval (CI): 45–78%) at false positive rate (FPR) of 0% (CI: 0–14%); good outcome could be predicted at 12 h with a sensitivity of 50% (CI: 29–71%) at a FPR of 5% (CI: 1–18%). Conclusion Deep learning of raw EEG signals outperforms any previously reported outcome predictor of coma after cardiac arrest, including visual assessment by trained EEG expert. Our approach offers the potential for objective and real-time insight in the prognosis of neurological outcome on a continuous scale, and can provide low-cost expertise at the bedside.