On-demand EEG education through competition - A novel, app-based approach to learning to identify interictal epileptiform discharges

Objective

Misinterpretation of EEGs harms patients, yet few resources exist to help trainees practice interpreting EEGs. We therefore sought to evaluate a novel educational tool to teach trainees how to identify interictal epileptiform discharges (IEDs) on EEG.

Methods

We created a public EEG test within the iOS app DiagnosUs using a pool of 13,262 candidate IEDs. Users were shown a candidate IED on EEG and asked to rate it as epileptiform (IED) or not (non-IED). They were given immediate feedback based on a gold standard. Learning was analyzed using a parametric model. We additionally analyzed IED features that best correlated with expert ratings.

Results

Our analysis included 901 participants. Users achieved a mean improvement of 13% over 1,000 questions and an ending accuracy of 81%. Users and experts appeared to rely on a similar set of IED morphologic features when analyzing candidate IEDs. We additionally identified particular types of candidate EEGs that remained challenging for most users even after substantial practice.

Conclusions

Users improved in their ability to properly classify candidate IEDs through repeated exposure and immediate feedback.

Significance

This app-based learning activity has great potential to be an effective supplemental tool to teach neurology trainees how to accurately identify IEDs on EEG.

Measuring expertise in identifying interictal epileptiform discharges

OBJECTIVE

Interictal epileptiform discharges on EEG are integral to diagnosing epilepsy. However, EEGs are interpreted by readers with and without specialty training, and there is no accepted method to assess skill in interpretation. We aimed to develop a test to quantify IED recognition skills.

METHODS

A total of 13,262 candidate IEDs were selected from EEGs and scored by eight fellowship-trained reviewers to establish a gold standard. An online test was developed to assess how well readers with different training levels could distinguish candidate waveforms. Sensitivity, false positive rate and calibration were calculated for each reader. A simple mathematical model was developed to estimate each reader's skill and threshold in identifying an IED, and to develop receiver operating characteristics curves for each reader. We investigated the number of IEDs needed to measure skill level with acceptable precision.

RESULTS

Twenty-nine raters completed the test; nine experts, seven experienced non-experts and thirteen novices. Median calibration errors for experts, experienced non-experts and novices were -0.056, 0.012, 0.046; median sensitivities were 0.800, 0.811, 0.715; and median false positive rates were 0.177, 0.272, 0.396, respectively. The number of test questions needed to measure those scores was 549. Our analysis identified that novices had a higher noise level (uncertainty) compared to experienced non-experts and experts. Using calculated noise and threshold levels, receiver operating curves were created, showing increasing median area under the curve from novices (0.735), to experienced non-experts (0.852) and experts (0.891).

SIGNIFICANCE

Expert and non-expert readers can be distinguished based on ability to identify IEDs. This type of assessment could also be used to identify and correct differences in thresholds in identifying IEDs.

Microscale dynamics of electrophysiological markers of epilepsy

OBJECTIVE

Interictal discharges (IIDs) and high frequency oscillations (HFOs) are established neurophysiologic biomarkers of epilepsy, while microseizures are less well studied. We used custom poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) microelectrodes to better understand these markers' microscale spatial dynamics.

METHODS

Electrodes with spatial resolution down to 50 µm were used to record intraoperatively in 30 subjects. IIDs' degree of spread and spatiotemporal paths were generated by peak-tracking followed by clustering. Repeating HFO patterns were delineated by clustering similar time windows. Multi-unit activity (MUA) was analyzed in relation to IID and HFO timing.

RESULTS

We detected IIDs encompassing the entire array in 93% of subjects, while localized IIDs, observed across < 50% of channels, were seen in 53%. IIDs traveled along specific paths. HFOs appeared in small, repeated spatiotemporal patterns. Finally, we identified microseizure events that spanned 50-100 µm. HFOs covaried with MUA, but not with IIDs.

CONCLUSIONS

Overall, these data suggest that irritable cortex micro-domains may form part of an underlying pathologic architecture which could contribute to the seizure network.

SIGNIFICANCE

These results, supporting the possibility that epileptogenic cortex comprises a mosaic of irritable domains, suggests that microscale approaches might be an important perspective in devising novel seizure control therapies.

Microscale Physiological Events on the Human Cortical Surface

Despite ongoing advances in our understanding of local single-cellular and network-level activity of neuronal populations in the human brain, extraordinarily little is known about their "intermediate" microscale local circuit dynamics. Here, we utilized ultra-high-density microelectrode arrays and a rare opportunity to perform intracranial recordings across multiple cortical areas in human participants to discover three distinct classes of cortical activity that are not locked to ongoing natural brain rhythmic activity. The first included fast waveforms similar to extracellular single-unit activity. The other two types were discrete events with slower waveform dynamics and were found preferentially in upper cortical layers. These second and third types were also observed in rodents, nonhuman primates, and semi-chronic recordings from humans via laminar and Utah array microelectrodes. The rates of all three events were selectively modulated by auditory and electrical stimuli, pharmacological manipulation, and cold saline application and had small causal co-occurrences. These results suggest that the proper combination of high-resolution microelectrodes and analytic techniques can capture neuronal dynamics that lay between somatic action potentials and aggregate population activity. Understanding intermediate microscale dynamics in relation to single-cell and network dynamics may reveal important details about activity in the full cortical circuit.

Development of Expert-Level Automated Detection of Epileptiform Discharges During Electroencephalogram Interpretation

Importance

Interictal epileptiform discharges (IEDs) in electroencephalograms (EEGs) are a biomarker of epilepsy, seizure risk, and clinical decline. However, there is a scarcity of experts qualified to interpret EEG results. Prior attempts to automate IED detection have been limited by small samples and have not demonstrated expert-level performance. There is a need for a validated automated method to detect IEDs with expert-level reliability.

Objective

To develop and validate a computer algorithm with the ability to identify IEDs as reliably as experts and classify an EEG recording as containing IEDs vs no IEDs.

Design, Setting, and Participants

A total of 9571 scalp EEG records with and without IEDs were used to train a deep neural network (SpikeNet) to perform IED detection. Independent training and testing data sets were generated from 13 262 IED candidates, independently annotated by 8 fellowship-trained clinical neurophysiologists, and 8520 EEG records containing no IEDs based on clinical EEG reports. Using the estimated spike probability, a classifier designating the whole EEG recording as positive or negative was also built.

Main Outcomes and Measures

SpikeNet accuracy, sensitivity, and specificity compared with fellowship-trained neurophysiology experts for identifying IEDs and classifying EEGs as positive or negative or negative for IEDs. Statistical performance was assessed via calibration error and area under the receiver operating characteristic curve (AUC). All performance statistics were estimated using 10-fold cross-validation.

Results

SpikeNet surpassed both expert interpretation and an industry standard commercial IED detector, based on calibration error (SpikeNet, 0.041; 95% CI, 0.033-0.049; vs industry standard, 0.066; 95% CI, 0.060-0.078; vs experts, mean, 0.183; range, 0.081-0.364) and binary classification performance based on AUC (SpikeNet, 0.980; 95% CI, 0.977-0.984; vs industry standard, 0.882; 95% CI, 0.872-0.893). Whole EEG classification had a mean calibration error of 0.126 (range, 0.109-0.1444) vs experts (mean, 0.197; range, 0.099-0.372) and AUC of 0.847 (95% CI, 0.830-0.865).

Conclusions and Relevance

In this study, SpikeNet automatically detected IEDs and classified whole EEGs as IED-positive or IED-negative. This may be the first time an algorithm has been shown to exceed expert performance for IED detection in a representative sample of EEGs and may thus be a valuable tool for expedited review of EEGs.

Interrater Reliability of Experts in Identifying Interictal Epileptiform Discharges in Electroencephalograms

Importance

The validity of using electroencephalograms (EEGs) to diagnose epilepsy requires reliable detection of interictal epileptiform discharges (IEDs). Prior interrater reliability (IRR) studies are limited by small samples and selection bias.

Objective

To assess the reliability of experts in detecting IEDs in routine EEGs.

Design, Setting, and Participants

This prospective analysis conducted in 2 phases included as participants physicians with at least 1 year of subspecialty training in clinical neurophysiology. In phase 1, 9 experts independently identified candidate IEDs in 991 EEGs (1 expert per EEG) reported in the medical record to contain at least 1 IED, yielding 87 636 candidate IEDs. In phase 2, the candidate IEDs were clustered into groups with distinct morphological features, yielding 12 602 clusters, and a representative candidate IED was selected from each cluster. We added 660 waveforms (11 random samples each from 60 randomly selected EEGs reported as being free of IEDs) as negative controls. Eight experts independently scored all 13 262 candidates as IEDs or non-IEDs. The 1051 EEGs in the study were recorded at the Massachusetts General Hospital between 2012 and 2016.

Main Outcomes and Measures

Primary outcome measures were percentage of agreement (PA) and beyond-chance agreement (Gwet κ) for individual IEDs (IED-wise IRR) and for whether an EEG contained any IEDs (EEG-wise IRR). Secondary outcomes were the correlations between numbers of IEDs marked by experts across cases, calibration of expert scoring to group consensus, and receiver operating characteristic analysis of how well multivariate logistic regression models may account for differences in the IED scoring behavior between experts.

Results

Among the 1051 EEGs assessed in the study, 540 (51.4%) were those of females and 511 (48.6%) were those of males. In phase 1, 9 experts each marked potential IEDs in a median of 65 (interquartile range [IQR], 28-332) EEGs. The total number of IED candidates marked was 87 636. Expert IRR for the 13 262 individually annotated IED candidates was fair, with the mean PA being 72.4% (95% CI, 67.0%-77.8%) and mean κ being 48.7% (95% CI, 37.3%-60.1%). The EEG-wise IRR was substantial, with the mean PA being 80.9% (95% CI, 76.2%-85.7%) and mean κ being 69.4% (95% CI, 60.3%-78.5%). A statistical model based on waveform morphological features, when provided with individualized thresholds, explained the median binary scores of all experts with a high degree of accuracy of 80% (range, 73%-88%).

Conclusions and Relevance

This study's findings suggest that experts can identify whether EEGs contain IEDs with substantial reliability. Lower reliability regarding individual IEDs may be largely explained by various experts applying different thresholds to a common underlying statistical model.

Receptor tyrosine kinase gene amplification is predictive of intraoperative seizures during glioma resection with functional mapping

OBJECTIVE

Intraoperative seizures during craniotomy with functional mapping is a common complication that impedes optimal tumor resection and results in significant morbidity. The relationship between genetic mutations in gliomas and the incidence of intraoperative seizures has not been well characterized. Here, the authors performed a retrospective study of patients treated at their institution over the last 12 years to determine whether molecular data can be used to predict the incidence of this complication.

METHODS

The authors queried their institutional database for patients with brain tumors who underwent resection with intraoperative functional mapping between 2005 and 2017. Basic clinicopathological characteristics, including the status of the following genes, were recorded: IDH1/2, PIK3CA, BRAF, KRAS, AKT1, EGFR, PDGFRA, MET, MGMT, and 1p/19q. Relationships between gene alterations and intraoperative seizures were evaluated using chi-square and two-sample t-test univariate analysis. When considering multiple predictive factors, a logistic multivariate approach was taken.

RESULTS

Overall, 416 patients met criteria for inclusion; of these patients, 98 (24%) experienced an intraoperative seizure. Patients with a history of preoperative seizure and those treated with antiepileptic drugs prior to surgery were less likely to have intraoperative seizures (history: OR 0.61 [95% CI 0.38-0.96], chi-square = 4.65, p = 0.03; AED load: OR 0.46 [95% CI 0.26-0.80], chi-square = 7.64, p = 0.01). In a univariate analysis of genetic markers, amplification of genes encoding receptor tyrosine kinases (RTKs) was specifically identified as a positive predictor of seizures (OR 5.47 [95% CI 1.22-24.47], chi-square = 5.98, p = 0.01). In multivariate analyses considering RTK status, AED use, and either 2007 WHO tumor grade or modern 2016 WHO tumor groups, the authors found that amplification of the RTK proto-oncogene, MET, was most predictive of intraoperative seizure (p < 0.05).

CONCLUSIONS

This study describes a previously unreported association between genetic alterations in RTKs and the occurrence of intraoperative seizures during glioma resection with functional mapping. Future models estimating intraoperative seizure risk may be enhanced by inclusion of genetic criteria.

SCOPE-mTL: A non-invasive tool for identifying and lateralizing mesial temporal lobe seizures prior to scalp EEG ictal onset

OBJECTIVE

In mesial temporal lobe (mTL) epilepsy, seizure onset can precede the appearance of a scalp EEG ictal pattern by many seconds. The ability to identify this early, occult mTL seizure activity could improve lateralization and localization of mTL seizures on scalp EEG.

METHODS

Using scalp EEG spectral features and machine learning approaches on a dataset of combined scalp EEG and foramen ovale electrode recordings in patients with mTL epilepsy, we developed an algorithm, SCOPE-mTL, to detect and lateralize early, occult mTL seizure activity, prior to the appearance of a scalp EEG ictal pattern.

RESULTS

Using SCOPE-mTL, 73% of seizures with occult mTL onset were identified as such, and no seizures that lacked an occult mTL onset were identified as having one. Predicted mTL seizure onset times were highly correlated with actual mTL seizure onset times (r=0.69). 50% of seizures with early mTL onset were lateralizable prior to scalp ictal onset, with 94% accuracy.

CONCLUSIONS

SCOPE-mTL can identify and lateralize mTL seizures prior to scalp EEG ictal onset, with high sensitivity, specificity, and accuracy.

SIGNIFICANCE

Quantitative analysis of scalp EEG can provide important information about mTL seizures, even in the absence of a visible scalp EEG ictal correlate.

EEG interpretation reliability and interpreter confidence: a large single-center study

The intrarater and interrater reliability (I&IR) of EEG interpretation has significant implications for the value of EEG as a diagnostic tool. We measured both the intrarater reliability and the interrater reliability of EEG interpretation based on the interpretation of complete EEGs into standard diagnostic categories and rater confidence in their interpretations and investigated sources of variance in EEG interpretations. During two distinct time intervals, six board-certified clinical neurophysiologists classified 300 EEGs into one or more of seven diagnostic categories and assigned a subjective confidence to their interpretations. Each EEG was read by three readers. Each reader interpreted 150 unique studies, and 50 studies were re-interpreted to generate intrarater data. A generalizability study assessed the contribution of subjects, readers, and the interaction between subjects and readers to interpretation variance. Five of the six readers had a median confidence of ≥99%, and the upper quartile of confidence values was 100% for all six readers. Intrarater Cohen's kappa (κc) ranged from 0.33 to 0.73 with an aggregated value of 0.59. Cohen's kappa ranged from 0.29 to 0.62 for the 15 reader pairs, with an aggregated Fleiss kappa of 0.44 for interrater agreement. Cohen's kappa was not significantly different across rater pairs (chi-square=17.3, df=14, p=0.24). Variance due to subjects (i.e., EEGs) was 65.3%, due to readers was 3.9%, and due to the interaction between readers and subjects was 30.8%. Experienced epileptologists have very high confidence in their EEG interpretations and low to moderate I&IR, a common paradox in clinical medicine. A necessary, but insufficient, condition to improve EEG interpretation accuracy is to increase intrarater and interrater reliability. This goal could be accomplished, for instance, with an automated online application integrated into a continuing medical education module that measures and reports EEG I&IR to individual users.

Mapping and mining interictal pathological gamma (30-100 Hz) oscillations with clinical intracranial EEG in patients with epilepsy

Localizing an epileptic network is essential for guiding neurosurgery and antiepileptic medical devices as well as elucidating mechanisms that may explain seizure-generation and epilepsy. There is increasing evidence that pathological oscillations may be specific to diseased networks in patients with epilepsy and that these oscillations may be a key biomarker for generating and indentifying epileptic networks. We present a semi-automated method that detects, maps, and mines pathological gamma (30-100 Hz) oscillations (PGOs) in human epileptic brain to possibly localize epileptic networks. We apply the method to standard clinical iEEG (<100 Hz) with interictal PGOs and seizures from six patients with medically refractory epilepsy. We demonstrate that electrodes with consistent PGO discharges do not always coincide with clinically determined seizure onset zone (SOZ) electrodes but at times PGO-dense electrodes include secondary seizure-areas (SS) or even areas without seizures (NS). In 4/5 patients with epilepsy surgery, we observe poor (Engel Class 4) post-surgical outcomes and identify more PGO-activity in SS or NS than in SOZ. Additional studies are needed to further clarify the role of PGOs in epileptic brain.

Data mining neocortical high-frequency oscillations in epilepsy and controls

Transient high-frequency (100-500 Hz) oscillations of the local field potential have been studied extensively in human mesial temporal lobe. Previous studies report that both ripple (100-250 Hz) and fast ripple (250-500 Hz) oscillations are increased in the seizure-onset zone of patients with mesial temporal lobe epilepsy. Comparatively little is known, however, about their spatial distribution with respect to seizure-onset zone in neocortical epilepsy, or their prevalence in normal brain. We present a quantitative analysis of high-frequency oscillations and their rates of occurrence in a group of nine patients with neocortical epilepsy and two control patients with no history of seizures. Oscillations were automatically detected and classified using an unsupervised approach in a data set of unprecedented volume in epilepsy research, over 12 terabytes of continuous long-term micro- and macro-electrode intracranial recordings, without human preprocessing, enabling selection-bias-free estimates of oscillation rates. There are three main results: (i) a cluster of ripple frequency oscillations with median spectral centroid = 137 Hz is increased in the seizure-onset zone more frequently than a cluster of fast ripple frequency oscillations (median spectral centroid = 305 Hz); (ii) we found no difference in the rates of high frequency oscillations in control neocortex and the non-seizure-onset zone neocortex of patients with epilepsy, despite the possibility of different underlying mechanisms of generation; and (iii) while previous studies have demonstrated that oscillations recorded by parenchyma-penetrating micro-electrodes have higher peak 100-500 Hz frequencies than penetrating macro-electrodes, this was not found for the epipial electrodes used here to record from the neocortical surface. We conclude that the relative rate of ripple frequency oscillations is a potential biomarker for epileptic neocortex, but that larger prospective studies correlating high-frequency oscillations rates with seizure-onset zone, resected tissue and surgical outcome are required to determine the true predictive value.

A neurological cause of recurrent choking during sleep

We describe a case of nocturnal choking episodes caused by insular seizures. Recurrent choking spells from sleep showed no response to treatment for sleep apnea or gastroesophageal reflux. Laryngoscopy revealed no abnormalities. Although continuous EEG monitoring during events was normal, ictal SPECT imaging showed increased radiotracer uptake in the left insular region, an area involved in sensation of the upper gastrointestinal tract. The episodes remitted after initiation of an antiepileptic drug. Obstructive sleep apnea is the most common cause for presentation to a sleep center, but seizures should remain in the differential diagnosis of nocturnal choking episodes.

Is the ABCD Score Useful for Risk Stratification of Patients With Acute Transient Ischemic Attack?

Background and Purpose— A 6-point scoring system (ABCD) was described recently for stratifying risk after transient ischemic attack (TIA). This score incorporates age (A), blood pressure (B), clinical features (C), and duration (D) of TIA. A score <4 reportedly indicates minimal short-term stroke risk. We evaluated this scoring system in an independent population.

Methods— This was a prospective study of TIA patients (diagnosed by a neurologist using the classic <24-hour definition) hospitalized <48 hours from symptom onset. The primary outcome assessment consisted of dichotomization of patients into 2 groups. The high-risk group included patients with stroke or death within 90 days, ≥50% stenosis in a relevant artery, or a cardioembolic source warranting anticoagulation. All others were classified as low risk. Findings on diffusion-weighted MRI (DWI) were also evaluated when performed and patients classified as DWI+ or DWI−.

Results— Over 3 years, 117 patients were enrolled. Median time from symptom onset to enrollment was 25.2 hours (interquartile range 19.8 to 30.2). Overall, 26 patients (22%) were classified as high risk, including 2 strokes, 2 deaths, 15 with ≥50% stenosis, and 10 with cardioembolic source. The frequency of high-risk patients increased with ABCD score (0 to 1 13%; 2 8%; 3 17%; 4 27%; 5 26%; 6 30%; P for trend=0.11). ABCD scores in the 2 patients with stroke were 3 and 6. Of those who underwent MRI, 15 of 61 (25%) were DWI+, but this correlated poorly with ABCD score (0 to 1 17%; 2 10%; 3 36%; 4 24%; 5 13%; 6 60%; P for trend=0.24).

Conclusions— Although the ABCD score has some predictive value, patients with a score <4 still have a substantial probability of having a high-risk cause of cerebral ischemia or radiographic evidence of acute infarction despite transient symptoms.