• Corpus ID: 7952067

Deep Architectures for Automated Seizure Detection in Scalp EEGs

@article{Golmohammadi2017DeepAF,
  title={Deep Architectures for Automated Seizure Detection in Scalp EEGs},
  author={Meysam Golmohammadi and Saeedeh Ziyabari and Vinit Shah and Silvia Lopez de Diego and Iyad Obeid and Joseph W. Picone},
  journal={ArXiv},
  year={2017},
  volume={abs/1712.09776}
}
Automated seizure detection using clinical electroencephalograms is a challenging machine learning problem because the multichannel signal often has an extremely low signal to noise ratio. Events of interest such as seizures are easily confused with signal artifacts (e.g, eye movements) or benign variants (e.g., slowing). Commercially available systems suffer from unacceptably high false alarm rates. Deep learning algorithms that employ high dimensional models have not previously been effective… 

Figures and Tables from this paper

Optimizing channel selection for seizure detection

The performance of a deep learning algorithm, CNN-LSTM, on several channel configurations designed to minimize the amount of spatial information lost is investigated, highlighting the importance of retaining referential channels for artifact reduction.

Temporal Graph Convolutional Networks for Automatic Seizure Detection

The temporal graph convolutional network (TGCN) is proposed, a model that leverages structural information and has relatively few parameters that provides a useful inductive bias in tasks where one expects similar features to be discriminative across the different sequences.

A Deep Convolutional Neural Network Method to Detect Seizures and Characteristic Frequencies Using Epileptic Electroencephalogram (EEG) Data

This is the first study to compute the contribution of frequency components to target seizure classification; thus allowing the identification of distinct seizure-related EEG frequency components compared to normal EEG measures.

EEG-Based Epileptic Seizure Detection via Machine/Deep Learning Approaches: A Systematic Review

An extensive systematic review on the recent developments of EEG-based ML/DL technologies for epileptic seizure diagnosis is conducted to aid researchers in deciding the most efficient ML/ DL models with optimal feature extraction methods to improve the performance of EEG's low amplitude and nonstationary characteristics.

Automatic Seizure Detection Using Modified CNN Architecture and Activation Layer

Inference based on evaluation can be drawn as the proposed hybrid architecture (AG86) showed better test results compared to pre-trained CNN models, and by replacing ReLU with the SWISH activation function, the performance of AlexNet and GoogLeNet improved.

DWT-Net: Seizure Detection System with Structured EEG Montage and Multiple Feature Extractor in Convolution Neural Network

A neural network model is proposed to extract features from EEG signals with a method of arranging the dimension of feature extraction inspired by the traditional method of neurologists, which approaches the performance of average human detector based on qEEG tools.

Gated recurrent networks for seizure detection

This study compares two types of recurrent units: long short-term memory units (LSTM) and gated recurrent units (GRU) using a state of the art hybrid architecture that integrates Convolutional Neural Networks (CNNs) with RNNs and explores regularization of these convolutional gating recurrent networks to address the problem of overfitting.

A Deep Learning Approach to Phase-Space Analysis for Seizure Detection

A method for epileptic seizure detection based on nonlinear phase space analysis and deep convolutional neural networks (CNN) and the underlying dynamics of scalp electroencephalography (sEEG) are extracted through time delay embedding and phase-space reconstruction.

References

SHOWING 1-10 OF 35 REFERENCES

EEG seizure detection and prediction algorithms: a survey

This paper covers some of the state-of-the-art seizure detection and prediction algorithms and provides comparison between these algorithms and concludes with future research directions and open problems in this topic.

The Temple University Hospital EEG Data Corpus

A new corpus of EEG data is described, the TUH-EEG Corpus, which is an ongoing data collection effort that has recently released 14 years of clinical EEG data collected at Temple University Hospital and contains data from 22 subjects, mostly pediatric.

Improved EEG event classification using differential energy

A comparison of a variety of approaches to estimating and postprocessing features on the TUH EEG Corpus, and a variant of a standard filter bank-based approach, coupled with first and second derivatives, provides a substantial reduction in the overall error rate.

Seizure detection, seizure prediction, and closed-loop warning systems in epilepsy

Automatic recognition of epileptic seizures in the EEG.

  • J. Gotman
  • Medicine
    Electroencephalography and clinical neurophysiology
  • 1982

Seizure detection: correlation of human experts

An analysis of two common reference points for EEGS

This investigation explores the statistical differences present in two different referential montages: Linked Ear (LE) and Averaged Reference (AR), and shows that a system trained on LE data significantly outperforms one trained only on AR data, suggesting that the impact of these statistical differences is subtler.

Diagnostic Accuracy of Electrographic Seizure Detection by Neurophysiologists and Non-Neurophysiologists in the Adult ICU Using a Panel of Quantitative EEG Trends

Quantitative EEG display panels are a promising tool to aid detection of seizures by non-neuroPhysiologists as well as by neurophysiologists, however, even when used as a panel, qEEG trends do not appear to be adequate as the sole method for reviewing continuous EEG data.

Dropout: a simple way to prevent neural networks from overfitting

It is shown that dropout improves the performance of neural networks on supervised learning tasks in vision, speech recognition, document classification and computational biology, obtaining state-of-the-art results on many benchmark data sets.