Epileptic seizure prediction using relative spectral power features

Bandarabadi, Mojtaba; Teixeira, César; Rasekhi, Jalil; Dourado, António

doi:10.1016/j.clinph.2014.05.022

Cited by 227 publications

(131 citation statements)

References 42 publications

Supporting

Mentioning

118

Contrasting

Unclassified

Order By: Relevance

“…The spectral power features are computationally very cost-effective, which makes them suitable candidates for implantable and portable warning systems. Furthermore, these features recently have demonstrated promising results for seizure prediction [15,35,39,[43][44][45][46], specifically in the high gamma frequency bands [15,[43][44][45][46]. Therefore, we investigated high-quality iEEG and sEEG recordings (with the sampling rates of 1024 Hz and 2500 Hz) and divided the gamma band into several narrower subbands to study the spectral behavior more precisely.…”

Section: Studied Featuresmentioning

confidence: 99%

“…Therefore, we investigated high-quality iEEG and sEEG recordings (with the sampling rates of 1024 Hz and 2500 Hz) and divided the gamma band into several narrower subbands to study the spectral behavior more precisely. Instead of using the well-known frequency bands, the subbands were selected as (0.5-4], (4)(5)(6)(7)(8), (8)(9)(10)(11)(12)(13)(14)(15), (15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30), (30)(31)(32)(33)(34)(35)(36)(37)(38)(39)(40)(41)(42)(43)(44)(45)(46)(47)(48) The spectral powers were obtained using power spectral density (PSD) estimated through Welch's method [47]. The PSD ...…”

Section: Studied Featuresmentioning

confidence: 99%

“…For some seizures, the epileptogenic transient changes develop very late close to the onset, whereas for some others, the epileptogenic transient changes appear much earlier several tens of minutes prior to the onset. Recently, several studies were proposed for seizure prediction using the combination of different univariate/bivariate and linear/nonlinear features, which leads to higher performances compared with using a single measure [10,14,15,[35][36][37][38][39]. For this purpose, we need to use machine learning approaches, which are mostly supervised.…”

Section: Introductionmentioning

confidence: 99%

“…However, as there exists no straightforward method for finding the appropriate preictal intervals of different seizures during the training phase, previous machine learning-based approaches have considered a fixed preictal period or examined different preictal periods in order to find a proper interval for labeling the training feature samples [10,14,15,[35][36][37][38][39]. In our previous studies [35,37,39,41], we investigated four different preictal periods of 10 min, 20 min, 30 min, and 40 min and trained classifiers to find the one providing the best results. Mormann et al [40] used statistical analysis of several univariate and bivariate features to prove the existence of such a transient preictal period.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

On the proper selection of preictal period for seizure prediction

Bandarabadi

Rasekhi

Teixeira

et al. 2015

Epilepsy & Behavior

Self Cite

View full text Add to dashboard Cite

Section: Studied Featuresmentioning

confidence: 99%

Section: Studied Featuresmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

On the proper selection of preictal period for seizure prediction

Bandarabadi

Rasekhi

Teixeira

et al. 2015

Epilepsy & Behavior

Self Cite

View full text Add to dashboard Cite

“…We were able to recalculate PPV for another study, which used long-term recordings from 24 human patients [20]. The mean PPV value across patients in that study was 31%.…”

Section: Discussionmentioning

confidence: 99%

Towards Improved Design and Evaluation of Epileptic Seizure Predictors

Korshunova

Kindermans

Degrave

et al. 2018

IEEE Trans. Biomed. Eng.

View full text Add to dashboard Cite

Abstract-Objective: Key issues in the epilepsy seizure prediction research are (1) the reproducibility of results (2) the inability to compare multiple approaches directly. To overcome these problems, the Seizure Prediction Challenge was organized on Kaggle.com. It aimed at establishing benchmarks on a dataset with predefined train, validation and test sets. Our main objective is to analyse the competition format, and to propose improvements, which would facilitate a better comparison of algorithms. The second objective is to present a novel deep learning approach to seizure prediction and compare it to other commonly used methods using patient centered metrics. Methods: We used the competition's datasets to illustrate the effects of data contamination. Having better data partitions, we compared three types of models in terms of different objectives. Results: We found that correct selection of test samples is crucial when evaluating the performance of seizure forecasting models. Moreover, we showed that models, which achieve state-of-the-art performance with respect to commonly used AUC, sensitivity and specificity metrics, may not yet be suitable for practical usage because of low precision scores. Conclusion: Correlation between validation and test datasets used in the competition limited its scientific value. Significance: Our findings provide guidelines which allow for a more objective evaluation of seizure prediction models.

show abstract

Deep learning‐based seizure prediction using EEG signals: A comparative analysis of classification methods on the CHB‐MIT dataset

Esmaeilpour,

Tabarestani,

Niazi

2024

Engineering Reports

View full text Add to dashboard Cite

Epilepsy is a brain disorder that causes patients to have multiple seizures. About 30% of patients with epilepsy are not treated with medication or surgery. The abnormal activity of brain before occurring of a seizure (about a few minutes before a seizure occurs) are known as the preictal area. Therefore, if we can predict this state, we can control possible seizures by using appropriate medications. In this study, we present a method for predicting epileptic seizures using electroencephalogram (EEG) signals. The method can identify the preictal region that occurs before the onset of seizures. In our proposed method, first the noise removal of EEG signals is performed, and then the necessary features are extracted using a convolution neural network. Finally, we use the feature vectors in order to train multiple classifiers, fully connected layer, random forest, and support vector machines with linear kernel. Additionally, we apply maximum voting, which is an ensemble method, to classify preictal segments from interictal ones. In this study, using EEG signals of patients from CHB‐MIT dataset, we were able to achieve sensitivity of 90.76%.

show abstract

Epileptic seizure prediction using relative spectral power features

Cited by 227 publications

References 42 publications

On the proper selection of preictal period for seizure prediction

On the proper selection of preictal period for seizure prediction

Towards Improved Design and Evaluation of Epileptic Seizure Predictors

Deep learning‐based seizure prediction using EEG signals: A comparative analysis of classification methods on the CHB‐MIT dataset

Contact Info

Product

Resources

About