High frequency oscillations in human memory and cognition: a neurophysiological substrate of engrams?

Despite advances in understanding the cellular and molecular processes underlying memory and cognition, and recent successful modulation of cognitive performance in brain disorders, the neurophysiological mechanisms remain underexplored. High frequency oscillations beyond the classic electroencephalogram spectrum have emerged as a potential neural correlate of fundamental cognitive processes. High frequency oscillations are detected in the human mesial temporal lobe and neocortical intracranial recordings spanning gamma/epsilon (60-150 Hz), ripple (80-250 Hz) and higher frequency ranges. Separate from other non-oscillatory activities, these brief electrophysiological oscillations of distinct duration, frequency and amplitude are thought to be generated by coordinated spiking of neuronal ensembles within volumes as small as a single cortical column. Although the exact origins, mechanisms, and physiological roles in health and disease remain elusive, they have been associated with human memory consolidation and cognitive processing. Recent studies suggest their involvement in encoding and recall of episodic memory with a possible role in the formation and reactivation of memory traces. High frequency oscillations are detected during encoding, throughout maintenance, and right before recall of remembered items, meeting a basic definition for an engram activity. The temporal coordination of high frequency oscillations reactivated across cortical and subcortical neural networks is ideally suited for integrating multimodal memory representations, which can be replayed and consolidated during states of wakefulness and sleep. High frequency oscillations have been shown to reflect coordinated bursts of neuronal assembly firing and offer a promising substrate for tracking and modulation of the hypothetical electrophysiological engram.

Timing matters for accurate identification of the epileptogenic zone.

OBJECTIVE: Interictal biomarkers of the epileptogenic zone (EZ) and their use in machine learning models open promising avenues for improvement of epilepsy surgery evaluation. Currently, most studies restrict their analysis to short segments of intracranial EEG (iEEG). METHODS: We used 2381 hours of iEEG data from 25 patients to systematically select 5-minute segments across various interictal conditions. Then, we tested machine learning models for EZ localization using iEEG features calculated within these individual segments or across them and evaluated the performance by the area under the precision-recall curve (PRAUC). RESULTS: On average, models achieved a score of 0.421 (the result of the chance classifier was 0.062). However, the PRAUC varied significantly across the segments (0.323-0.493). Overall, NREM sleep achieved the highest scores, with the best results of 0.493 in N2. When using data from all segments, the model performed significantly better than single segments, except NREM sleep segments. CONCLUSIONS: The model based on a short segment of iEEG recording can achieve similar results as a model based on prolonged recordings. The analyzed segment should, however, be carefully and systematically selected, preferably from NREM sleep. SIGNIFICANCE: Random selection of short iEEG segments may give rise to inaccurate localization of the EZ.

Interictal invasive very high-frequency oscillations in resting awake state and sleep.

Interictal very high-frequency oscillations (VHFOs, 500-2000 Hz) in a resting awake state seem to be, according to a precedent study of our team, a more specific predictor of a good outcome of the epilepsy surgery compared to traditional interictal high-frequency oscillations (HFOs, 80-500 Hz). In this study, we retested this hypothesis on a larger cohort of patients. In addition, we also collected patients' sleep data and hypothesized that the occurrence of VHFOs in sleep will be greater than in resting state. We recorded interictal invasive electroencephalographic (iEEG) oscillations in 104 patients with drug-resistant epilepsy in a resting state and in 35 patients during sleep. 21 patients in the rest study and 11 patients in the sleep study met the inclusion criteria (interictal HFOs and VHFOs present in iEEG recordings, a surgical intervention and a postoperative follow-up of at least 1 year) for further evaluation of iEEG data. In the rest study, patients with good postoperative outcomes had significantly higher ratio of resected contacts with VHFOs compared to HFOs. In sleep, VHFOs were more abundant than in rest and the percentage of resected contacts in patients with good and poor outcomes did not considerably differ in any type of oscillations. In conclusion, (1) our results confirm, in a larger patient cohort, our previous work about VHFOs being a specific predictor of the area which needs to be resected; and (2) that more frequent sleep VHFOs do not further improve the results.

Interictal high-frequency oscillations, spikes, and connectivity profiles: A fingerprint of epileptogenic brain pathologies.

OBJECTIVE: Focal cortical dysplasia (FCD), hippocampal sclerosis (HS), nonspecific gliosis (NG), and normal tissue (NT) comprise the majority of histopathological results of surgically treated drug-resistant epilepsy patients. Epileptic spikes, high-frequency oscillations (HFOs), and connectivity measures are valuable biomarkers of epileptogenicity. The question remains whether they could also be utilized for preresective differentiation of the underlying brain pathology. This study explored spikes and HFOs together with functional connectivity in various epileptogenic pathologies. METHODS: Interictal awake stereoelectroencephalographic recordings of 33 patients with focal drug-resistant epilepsy with seizure-free postoperative outcomes were analyzed (15 FCD, 8 HS, 6 NT, and 4 NG). Interictal spikes and HFOs were automatically identified in the channels contained in the overlap of seizure onset zone and resected tissue. Functional connectivity measures (relative entropy, linear correlation, cross-correlation, and phase consistency) were computed for neighboring electrode pairs. RESULTS: Statistically significant differences were found between the individual pathologies in HFO rates, spikes, and their characteristics, together with functional connectivity measures, with the highest values in the case of HS and NG/NT. A model to predict brain pathology based on all interictal measures achieved up to 84.0% prediction accuracy. SIGNIFICANCE: The electrophysiological profile of the various epileptogenic lesions in epilepsy surgery patients was analyzed. Based on this profile, a predictive model was developed. This model offers excellent potential to identify the nature of the underlying lesion prior to resection. If validated, this model may be particularly valuable for counseling patients, as depending on the lesion type, different outcomes are achieved after epilepsy surgery.

Relative entropy is an easy-to-use invasive electroencephalographic biomarker of the epileptogenic zone.

OBJECTIVE: High-frequency oscillations are considered among the most promising interictal biomarkers of the epileptogenic zone in patients suffering from pharmacoresistant focal epilepsy. However, there is no clear definition of pathological high-frequency oscillations, and the existing detectors vary in methodology, performance, and computational costs. This study proposes relative entropy as an easy-to-use novel interictal biomarker of the epileptic tissue. METHODS: We evaluated relative entropy and high-frequency oscillation biomarkers on intracranial electroencephalographic data from 39 patients with seizure-free postoperative outcome (Engel Ia) from three institutions. We tested their capability to localize the epileptogenic zone, defined as resected contacts located in the seizure onset zone. The performance was compared using areas under the receiver operating curves (AUROCs) and precision-recall curves. Then we tested whether a universal threshold can be used to delineate the epileptogenic zone across patients from different institutions. RESULTS: Relative entropy in the ripple band (80-250 Hz) achieved an average AUROC of .85. The normalized high-frequency oscillation rate in the ripple band showed an identical AUROC of .85. In contrast to high-frequency oscillations, relative entropy did not require any patient-level normalization and was easy and fast to calculate due to its clear and straightforward definition. One threshold could be set across different patients and institutions, because relative entropy is independent of signal amplitude and sampling frequency. SIGNIFICANCE: Although both relative entropy and high-frequency oscillations have a similar performance, relative entropy has significant advantages such as straightforward definition, computational speed, and universal interpatient threshold, making it an easy-to-use promising biomarker of the epileptogenic zone.

Shadows of very high-frequency oscillations can be detected in lower frequency bands of routine stereoelectroencephalography.

Very high-frequency oscillations (VHFOs, > 500 Hz) are more specific in localizing the epileptogenic zone (EZ) than high-frequency oscillations (HFOs, < 500 Hz). Unfortunately, VHFOs are not visible in standard clinical stereo-EEG (SEEG) recordings with sampling rates of 1 kHz or lower. Here we show that "shadows" of VHFOs can be found in frequencies below 500 Hz and can help us to identify SEEG channels with a higher probability of increased VHFO rates. Subsequent analysis of Logistic regression models on 141 SEEG channels from thirteen patients shows that VHFO "shadows" provide additional information to gold standard HFO analysis and can potentially help in precise EZ delineation in standard clinical recordings.

Distributed brain co-processor for tracking spikes, seizures and behaviour during electrical brain stimulation.

Early implantable epilepsy therapy devices provided open-loop electrical stimulation without brain sensing, computing, or an interface for synchronized behavioural inputs from patients. Recent epilepsy stimulation devices provide brain sensing but have not yet developed analytics for accurately tracking and quantifying behaviour and seizures. Here we describe a distributed brain co-processor providing an intuitive bi-directional interface between patient, implanted neural stimulation and sensing device, and local and distributed computing resources. Automated analysis of continuous streaming electrophysiology is synchronized with patient reports using a handheld device and integrated with distributed cloud computing resources for quantifying seizures, interictal epileptiform spikes and patient symptoms during therapeutic electrical brain stimulation. The classification algorithms for interictal epileptiform spikes and seizures were developed and parameterized using long-term ambulatory data from nine humans and eight canines with epilepsy, and then implemented prospectively in out-of-sample testing in two pet canines and four humans with drug-resistant epilepsy living in their natural environments. Accurate seizure diaries are needed as the primary clinical outcome measure of epilepsy therapy and to guide brain-stimulation optimization. The brain co-processor system described here enables tracking interictal epileptiform spikes, seizures and correlation with patient behavioural reports. In the future, correlation of spikes and seizures with behaviour will allow more detailed investigation of the clinical impact of spikes and seizures on patients.

Protocol for multicentre comparison of interictal high-frequency oscillations as a predictor of seizure freedom.

In drug-resistant focal epilepsy, interictal high-frequency oscillations (HFOs) recorded from intracranial EEG (iEEG) may provide clinical information for delineating epileptogenic brain tissue. The iEEG electrode contacts that contain HFO are hypothesized to delineate the epileptogenic zone; their resection should then lead to postsurgical seizure freedom. We test whether our prospective definition of clinically relevant HFO is in agreement with postsurgical seizure outcome. The algorithm is fully automated and is equally applied to all data sets. The aim is to assess the reliability of the proposed detector and analysis approach. We use an automated data-independent prospective definition of clinically relevant HFO that has been validated in data from two independent epilepsy centres. In this study, we combine retrospectively collected data sets from nine independent epilepsy centres. The analysis is blinded to clinical outcome. We use iEEG recordings during NREM sleep with a minimum of 12 epochs of 5 min of NREM sleep. We automatically detect HFO in the ripple (80-250 Hz) and in the fast ripple (250-500 Hz) band. There is no manual rejection of events in this fully automated algorithm. The type of HFO that we consider clinically relevant is defined as the simultaneous occurrence of a fast ripple and a ripple. We calculate the temporal consistency of each patient's HFO rates over several data epochs within and between nights. Patients with temporal consistency <50% are excluded from further analysis. We determine whether all electrode contacts with high HFO rate are included in the resection volume and whether seizure freedom (ILAE 1) was achieved at ≥2 years follow-up. Applying a previously validated algorithm to a large cohort from several independent epilepsy centres may advance the clinical relevance and the generalizability of HFO analysis as essential next step for use of HFO in clinical practice.

Intracranial electrophysiological recordings from the human brain during memory tasks with pupillometry.

Data comprise intracranial EEG (iEEG) brain activity represented by stereo EEG (sEEG) signals, recorded from over 100 electrode channels implanted in any one patient across various brain regions. The iEEG signals were recorded in epilepsy patients (N = 10) undergoing invasive monitoring and localization of seizures when they were performing a battery of four memory tasks lasting approx. 1 hour in total. Gaze tracking on the task computer screen with estimating the pupil size was also recorded together with behavioral performance. Each dataset comes from one patient with anatomical localization of each electrode contact. Metadata contains labels for the recording channels with behavioral events marked from all tasks, including timing of correct and incorrect vocalization of the remembered stimuli. The iEEG and the pupillometric signals are saved in BIDS data structure to facilitate efficient data sharing and analysis.

Ultra-fast oscillation detection in EEG signal from deep-brain microelectrodes.

For the last decades, ripples 80-200Hz (R)and fast ripples 200-500Hz (FR) were intensively studied as biomarkers of the epileptogenic zone (EZ). Recently, Very fast ripples 500-1000Hz (VFR) and ultra-fast ripples 1000-2000Hz (UFR) recorded using standard clinical macro electrodes have been shown to be more specific for EZ. High-sampled microelectrode recordings can bring new insights into this phenomenon of high frequency, multiunit activity. Unfortunately, visual detection of such events is extremely time consuming and unreliable. Here we present a detector of ultra-fast oscillations (UFO, >1kHz). In an example of two patients, we detected 951 UFOs which were more frequent in epileptic (8.6/min) vs. non-epileptic hippocampus (1.3/min). Our detection method can serve as a tool for exploring extremely high frequency events from microelectrode recordings.

Cognitive Processing Impacts High Frequency Intracranial EEG Activity of Human Hippocampus in Patients With Pharmacoresistant Focal Epilepsy.

The electrophysiological EEG features such as high frequency oscillations, spikes and functional connectivity are often used for delineation of epileptogenic tissue and study of the normal function of the brain. The epileptogenic activity is also known to be suppressed by cognitive processing. However, differences between epileptic and healthy brain behavior during rest and task were not studied in detail. In this study we investigate the impact of cognitive processing on epileptogenic and non-epileptogenic hippocampus and the intracranial EEG features representing the underlying electrophysiological processes. We investigated intracranial EEG in 24 epileptic and 24 non-epileptic hippocampi in patients with intractable focal epilepsy during a resting state period and during performance of various cognitive tasks. We evaluated the behavior of features derived from high frequency oscillations, interictal epileptiform discharges and functional connectivity and their changes in relation to cognitive processing. Subsequently, we performed an analysis whether cognitive processing can contribute to classification of epileptic and non-epileptic hippocampus using a machine learning approach. The results show that cognitive processing suppresses epileptogenic activity in epileptic hippocampus while it causes a shift toward higher frequencies in non-epileptic hippocampus. Statistical analysis reveals significantly different electrophysiological reactions of epileptic and non-epileptic hippocampus during cognitive processing, which can be measured by high frequency oscillations, interictal epileptiform discharges and functional connectivity. The calculated features showed high classification potential for epileptic hippocampus (AUC = 0.93). In conclusion, the differences between epileptic and non-epileptic hippocampus during cognitive processing bring new insight in delineation between pathological and physiological processes. Analysis of computed iEEG features in rest and task condition can improve the functional mapping during pre-surgical evaluation and provide additional guidance for distinguishing between epileptic and non-epileptic structure which is absolutely crucial for achieving the best possible outcome with as little side effects as possible.

Stereo-electroencephalography (SEEG) reference based on low-variance signals.

For a correct assessment of stereo-electroencephalographic (SEEG) recordings, a proper signal electrical reference is necessary. Such a reference might be physical or virtual. Physical reference can be noisy and a proper virtual reference calculation is often time-consuming. This paper uses the variance of the SEEG signals to calculate the reference from relatively low noise signals to reduce the contamination by distant sources, while maintaining negligible computing time.Ten patients with SEEG recordings were used in this study. 20-second long recordings from each patient, sampled at 5000 Hz, were used to calculate variances of SEEG signals and a low-variance (LV) subset of signals was selected for each patient. Consequently, 4 different reference signals were calculated using: 1) an average signal from WM contacts only (AVG_WM); 2) an average signal from LV contacts only (AVG_LV); 3) independent component analysis (ICA) method from WM contacts only (ICA_WM); and 4) ICA method from LV signals only (ICA_LV). Also, the original testing reference, an average signal from all SEEG contacts (AVG) was utilized. Finally, bipolar signals and average signals from anatomical structures were calculated and used to evaluate reference signals.91.7% of the WM SEEG contacts were found below the average variance. ICA_LV showed the best and AVG_WM the worst overall results. AVG_LV had the most positive impact on minimizing the mutual correlations between separate brain structures and correcting the outliers. The average processing time for ICA methods was 66.72 seconds and 0.7870 seconds for AVG methods (100 000 samples, 125.7±20.4 SEEG signals).Utilizing the LV data subset improves the reference signal. WM references are difficult to obtain and seem to be more susceptible to errors caused by low number of WM contacts in the dataset. ICA_LV can be considered as one of the best reference estimations, however the calculation is very demanding and time consuming. AVG_LV shows good and stable results, while it is based on a straightforward methodology and outstandingly fast calculation.

High frequency oscillations in epileptic and non-epileptic human hippocampus during a cognitive task.

Hippocampal high-frequency electrographic activity (HFOs) represents one of the major discoveries not only in epilepsy research but also in cognitive science over the past few decades. A fundamental challenge, however, has been the fact that physiological HFOs associated with normal brain function overlap in frequency with pathological HFOs. We investigated the impact of a cognitive task on HFOs with the aim of improving differentiation between epileptic and non-epileptic hippocampi in humans. Hippocampal activity was recorded with depth electrodes in 15 patients with focal epilepsy during a resting period and subsequently during a cognitive task. HFOs in ripple and fast ripple frequency ranges were evaluated in both conditions, and their rate, spectral entropy, relative amplitude and duration were compared in epileptic and non-epileptic hippocampi. The similarity of HFOs properties recorded at rest in epileptic and non-epileptic hippocampi suggests that they cannot be used alone to distinguish between hippocampi. However, both ripples and fast ripples were observed with higher rates, higher relative amplitudes and longer durations at rest as well as during a cognitive task in epileptic compared with non-epileptic hippocampi. Moreover, during a cognitive task, significant reductions of HFOs rates were found in epileptic hippocampi. These reductions were not observed in non-epileptic hippocampi. Our results indicate that although both hippocampi generate HFOs with similar features that probably reflect non-pathological phenomena, it is possible to differentiate between epileptic and non-epileptic hippocampi using a simple odd-ball task.

Multicenter intracranial EEG dataset for classification of graphoelements and artifactual signals.

EEG signal processing is a fundamental method for neurophysiology research and clinical neurology practice. Historically the classification of EEG into physiological, pathological, or artifacts has been performed by expert visual review of the recordings. However, the size of EEG data recordings is rapidly increasing with a trend for higher channel counts, greater sampling frequency, and longer recording duration and complete reliance on visual data review is not sustainable. In this study, we publicly share annotated intracranial EEG data clips from two institutions: Mayo Clinic, MN, USA and St. Anne's University Hospital Brno, Czech Republic. The dataset contains intracranial EEG that are labeled into three groups: physiological activity, pathological/epileptic activity, and artifactual signals. The dataset published here should support and facilitate training of generalized machine learning and digital signal processing methods for intracranial EEG and promote research reproducibility. Along with the data, we also propose a statistical method that is recommended for comparison of candidate classifier performance utilizing out-of-institution/out-of-patient testing.

NREM sleep is the state of vigilance that best identifies the epileptogenic zone in the interictal electroencephalogram.

OBJECTIVE: Interictal epileptiform anomalies such as epileptiform discharges or high-frequency oscillations show marked variations across the sleep-wake cycle. This study investigates which state of vigilance is the best to localize the epileptogenic zone (EZ) in interictal intracranial electroencephalography (EEG). METHODS: Thirty patients with drug-resistant epilepsy undergoing stereo-EEG (SEEG)/sleep recording and subsequent open surgery were included; 13 patients (43.3%) had good surgical outcome (Engel class I). Sleep was scored following standard criteria. Multiple features based on the interictal EEG (interictal epileptiform discharges, high-frequency oscillations, univariate and bivariate features) were used to train a support vector machine (SVM) model to classify SEEG contacts placed in the EZ. The performance of the algorithm was evaluated by the mean area under the receiver-operating characteristic (ROC) curves (AUCs) and positive predictive values (PPVs) across 10-minute sections of wake, non-rapid eye movement sleep (NREM) stages N2 and N3, REM sleep, and their combination. RESULTS: Highest AUCs were achieved in NREM sleep stages N2 and N3 compared to wakefulness and REM (P < .01). There was no improvement when using a combination of all four states (P > .05); the best performing features in the combined state were selected from NREM sleep. There were differences between good (Engel I) and poor (Engel II-IV) outcomes in PPV (P < .05) and AUC (P < .01) across all states. The SVM multifeature approach outperformed spikes and high-frequency oscillations (P < .01) and resulted in results similar to those of the seizure-onset zone (SOZ; P > .05). SIGNIFICANCE: Sleep improves the localization of the EZ with best identification obtained in NREM sleep stages N2 and N3. Results based on the multifeature classification in 10 minutes of NREM sleep were not different from the results achieved by the SOZ based on 12.7 days of seizure monitoring. This finding might ultimately result in a more time-efficient intracranial presurgical investigation of focal epilepsy.

Multi-feature localization of epileptic foci from interictal, intracranial EEG.

OBJECTIVE: When considering all patients with focal drug-resistant epilepsy, as high as 40-50% of patients suffer seizure recurrence after surgery. To achieve seizure freedom without side effects, accurate localization of the epileptogenic tissue is crucial before its resection. We investigate an automated, fast, objective mapping process that uses only interictal data. METHODS: We propose a novel approach based on multiple iEEG features, which are used to train a support vector machine (SVM) model for classification of iEEG electrodes as normal or pathologic using 30 min of inter-ictal recording. RESULTS: The tissue under the iEEG electrodes, classified as epileptogenic, was removed in 17/18 excellent outcome patients and was not entirely resected in 8/10 poor outcome patients. The overall best result was achieved in a subset of 9 excellent outcome patients with the area under the receiver operating curve = 0.95. CONCLUSION: SVM models combining multiple iEEG features show better performance than algorithms using a single iEEG marker. Multiple iEEG and connectivity features in presurgical evaluation could improve epileptogenic tissue localization, which may improve surgical outcome and minimize risk of side effects. SIGNIFICANCE: In this study, promising results were achieved in localization of epileptogenic regions by SVM models that combine multiple features from 30 min of inter-ictal iEEG recordings.

Hippocampal high frequency oscillations in unilateral and bilateral mesial temporal lobe epilepsy.

OBJECTIVE: The main aim of this study was to investigate the potential differences in terms of interictal high frequency oscillations (HFOs) between both hippocampi in unilateral (U-MTLE) and bilateral mesial temporal lobe epilepsy (B-MTLE). METHODS: Sixteen patients with MTLE underwent bilateral hippocampal depth electrode implantation as part of epilepsy surgery evaluation. Interictal HFOs were detected automatically. The analyses entail comparisons of the rates and spatial distributions of ripples and fast ripples (FR) in hippocampi and amygdalae, with respect to the eventual finding of hippocampal sclerosis (HS). RESULTS: In U-MTLE, higher ripple and FR rates were found in the hippocampi ipsilateral to the seizure onset than in the contralateral hippocampi. Non-epileptic hippocampi in U-MTLE were distinguished by significantly lower ripple rate than in the remaining analyzed hippocampi. There were not differences between the hippocampi in B-MTLE. In the hippocampi with proven HS, higher FR rates were observed in the ventral than in the dorsal parts. CONCLUSIONS: Non-epileptic hippocampi in U-MTLE demonstrated significantly lower ripple rates than those epileptic in U-MTLE and B-MTLE. SIGNIFICANCE: Low interictal HFO occurrence might be considered as a marker of the non-epileptic hippocampi in MTLE.

Deep-learning for seizure forecasting in canines with epilepsy.

OBJECTIVE: This paper introduces a fully automated, subject-specific deep-learning convolutional neural network (CNN) system for forecasting seizures using ambulatory intracranial EEG (iEEG). The system was tested on a hand-held device (Mayo Epilepsy Assist Device) in a pseudo-prospective mode using iEEG from four canines with naturally occurring epilepsy. APPROACH: The system was trained and tested on 75 seizures collected over 1608 d utilizing a genetic algorithm to optimize forecasting hyper-parameters (prediction horizon (PH), median filter window length, and probability threshold) for each subject-specific seizure forecasting model. The trained CNN models were deployed on a hand-held tablet computer and tested on testing iEEG datasets from four canines. The results from the iEEG testing datasets were compared with Monte Carlo simulations using a Poisson random predictor with equal time in warning to evaluate seizure forecasting performance. MAIN RESULTS: The results show the CNN models forecasted seizures at rates significantly above chance in all four dogs (p  < 0.01, with mean 0.79 sensitivity and 18% time in warning). The deep learning method presented here surpassed the performance of previously reported methods using computationally expensive features with standard machine learning methods like logistic regression and support vector machine classifiers. SIGNIFICANCE: Our findings principally support the feasibility of deploying trained CNN models on a hand-held computational device (Mayo Epilepsy Assist Device) that analyzes streaming iEEG data for real-time seizure forecasting.

Intracerebral EEG Artifact Identification Using Convolutional Neural Networks.

Manual and semi-automatic identification of artifacts and unwanted physiological signals in large intracerebral electroencephalographic (iEEG) recordings is time consuming and inaccurate. To date, unsupervised methods to accurately detect iEEG artifacts are not available. This study introduces a novel machine-learning approach for detection of artifacts in iEEG signals in clinically controlled conditions using convolutional neural networks (CNN) and benchmarks the method's performance against expert annotations. The method was trained and tested on data obtained from St Anne's University Hospital (Brno, Czech Republic) and validated on data from Mayo Clinic (Rochester, Minnesota, U.S.A). We show that the proposed technique can be used as a generalized model for iEEG artifact detection. Moreover, a transfer learning process might be used for retraining of the generalized version to form a data-specific model. The generalized model can be efficiently retrained for use with different EEG acquisition systems and noise environments. The generalized and specialized model F1 scores on the testing dataset were 0.81 and 0.96, respectively. The CNN model provides faster, more objective, and more reproducible iEEG artifact detection compared to manual approaches.