Dataset Information

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

ABSTRACT: 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.

SUBMITTER: Nejedly P

PROVIDER: S-EPMC7297990 | biostudies-literature | 2020 Jun

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

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

Nejedly Petr P Kremen Vaclav V Sladky Vladimir V Cimbalnik Jan J Klimes Petr P Plesinger Filip F Mivalt Filip F Travnicek Vojtech V Viscor Ivo I Pail Martin M Halamek Josef J Brinkmann Benjamin H BH Brazdil Milan M Jurak Pavel P Worrell Gregory G

Scientific data 20200616 1

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 p ...[more]

PMID: 32546753

Similar Datasets

Project description:BackgroundArtifacts contained in EEG recordings hamper both, the visual interpretation by experts as well as the algorithmic processing and analysis (e.g. for Brain-Computer Interfaces (BCI) or for Mental State Monitoring). While hand-optimized selection of source components derived from Independent Component Analysis (ICA) to clean EEG data is widespread, the field could greatly profit from automated solutions based on Machine Learning methods. Existing ICA-based removal strategies depend on explicit recordings of an individual's artifacts or have not been shown to reliably identify muscle artifacts.MethodsWe propose an automatic method for the classification of general artifactual source components. They are estimated by TDSEP, an ICA method that takes temporal correlations into account. The linear classifier is based on an optimized feature subset determined by a Linear Programming Machine (LPM). The subset is composed of features from the frequency-, the spatial- and temporal domain. A subject independent classifier was trained on 640 TDSEP components (reaction time (RT) study, n = 12) that were hand labeled by experts as artifactual or brain sources and tested on 1080 new components of RT data of the same study. Generalization was tested on new data from two studies (auditory Event Related Potential (ERP) paradigm, n = 18; motor imagery BCI paradigm, n = 80) that used data with different channel setups and from new subjects.ResultsBased on six features only, the optimized linear classifier performed on level with the inter-expert disagreement (<10% Mean Squared Error (MSE)) on the RT data. On data of the auditory ERP study, the same pre-calculated classifier generalized well and achieved 15% MSE. On data of the motor imagery paradigm, we demonstrate that the discriminant information used for BCI is preserved when removing up to 60% of the most artifactual source components.ConclusionsWe propose a universal and efficient classifier of ICA components for the subject independent removal of artifacts from EEG data. Based on linear methods, it is applicable for different electrode placements and supports the introspection of results. Trained on expert ratings of large data sets, it is not restricted to the detection of eye- and muscle artifacts. Its performance and generalization ability is demonstrated on data of different EEG studies.

Project description:Brain Computer-Interface (BCI) technology plays a considerable role in the control of rehabilitation or peripheral devices for stroke patients. This is particularly due to their inability to control such devices from their inherent physical limitations after such an attack. More often than not, the control of such devices exploits electroencephalogram (EEG) signals. Nonetheless, it is worth noting that the extraction of the features and the classification of the signals is non-trivial for a successful BCI system. The use of Transfer Learning (TL) has been demonstrated to be a powerful tool in the extraction of essential features. However, the employment of such a method towards BCI applications, particularly in regard to EEG signals, are somewhat limited. The present study aims to evaluate the effectiveness of different TL models in extracting features for the classification of wink-based EEG signals. The extracted features are classified by means of fine-tuned Random Forest (RF) classifier. The raw EEG signals are transformed into a scalogram image via Continuous Wavelet Transform (CWT) before it was fed into the TL models, namely InceptionV3, Inception ResNetV2, Xception and MobileNet. The dataset was divided into training, validation, and test datasets, respectively, via a stratified ratio of 60:20:20. The hyperparameters of the RF models were optimised through the grid search approach, in which the five-fold cross-validation technique was adopted. The optimised RF classifier performance was compared with the conventional TL-based CNN classifier performance. It was demonstrated from the study that the best TL model identified is the Inception ResNetV2 along with an optimised RF pipeline, as it was able to yield a classification accuracy of 100% on both the training and validation dataset. Therefore, it could be established from the study that a comparable classification efficacy is attainable via the Inception ResNetV2 with an optimised RF pipeline. It is envisaged that the implementation of the proposed architecture to a BCI system would potentially facilitate post-stroke patients to lead a better life quality.

Project description:Intracranial electroencephalography (iEEG) has been the mainstay of identifying the seizure onset zone (SOZ), a key diagnostic procedure in addition to neuroimaging when considering epilepsy surgery. In many patients, iEEG has been the basis for resective epilepsy surgery, to date still the most successful treatment for drug-resistant epilepsy. Intracranial EEG determines the location and resectability of the SOZ. Advances in recording and implantation of iEEG provide multiple options in the 21st century. This not only includes the choice between subdural electrodes (SDE) and stereoelectroencephalography (SEEG) but also includes the implantation and recordings from microelectrodes. Before iEEG implantation, especially in magnetic resonance imaging -negative epilepsy, a clear hypothesis for seizure generation and propagation should be based on noninvasive methods. Intracranial EEG implantation should be planned by a multidisciplinary team considering epileptic networks. Recordings from SDE and SEEG have both their advantages and disadvantages. Stereo-EEG seems to have a lower rate of complications that are clinically significant, but has limitations in spatial sampling of the cortical surface. Stereo-EEG can sample deeper areas of the brain including deep sulci and hard to reach areas such as the insula. To determine the epileptogenic zone, interictal and ictal information should be taken into consideration. Interictal spiking, low frequency slowing, as well as high frequency oscillations may inform about the epileptogenic zone. Ictally, high frequency onsets in the beta/gamma range are usually associated with the SOZ, but specialized recordings with combined macro and microelectrodes may in the future educate us about onset in higher frequency bands. Stimulation of intracranial electrodes triggering habitual seizures can assist in identifying the SOZ. Advanced computational methods such as determining the epileptogenicity index and similar measures may enhance standard clinical interpretation. Improved techniques to record and interpret iEEG may in the future lead to a greater proportion of patients being seizure free after epilepsy surgery.

Project description:ObjectivesSimultaneous intracranial EEG and functional MRI (icEEG-fMRI) can be used to map the haemodynamic (BOLD) changes associated with the generation of IEDs. Unlike scalp EEG-fMRI, in most patients who undergo icEEG-fMRI, IEDs recorded intracranially are numerous and show variability in terms of field amplitude and morphology. Therefore, visual marking can be highly subjective and time consuming. In this study, we applied an automated spike classification algorithm, Wave_clus (WC), to IEDs marked visually on icEEG data acquired during simultaneous fMRI acquisition. The motivation of this work is to determine whether using a potentially more consistent and unbiased automated approach can produce more biologically meaningful BOLD patterns compared to the BOLD patterns obtained based on the conventional, visual classification.MethodsWe analysed simultaneous icEEG-fMRI data from eight patients with severe drug resistant epilepsy, and who subsequently underwent resective surgery that resulted in a good outcome: confirmed epileptogenic zone (EZ). For each patient two fMRI analyses were performed: one based on the conventional visual IED classification and the other based on the automated classification. We used the concordance of the IED-related BOLD maps with the confirmed EZ as an indication of their biological meaning, which we compared for the automated and visual classifications for all IED originating in the EZ.ResultsAcross the group, the visual and automated classifications resulted in 32 and 24 EZ IED classes respectively, for which 75% vs 83% of the corresponding BOLD maps were concordant. At the single-subject level, the BOLD maps for the automated approach had greater concordance in four patients, and less concordance in one patient, compared to those obtained using the conventional visual classification, and equal concordance for three remaining patients. These differences did not reach statistical significance.ConclusionWe found automated IED classification on icEEG data recorded during fMRI to be feasible and to result in IED-related BOLD maps that may contain similar or greater biological meaning compared to the conventional approach in the majority of the cases studied. We anticipate that this approach will help to gain significant new insights into the brain networks associated with IEDs and in relation to postsurgical outcome.

Dataset Information

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

Publications

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

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets