Epileptic seizure identification in EEG signals using DWT, ANN and sequential window algorithm
A patient-specific novel systematic methodology is described in this study for automatic seizure detection from raw electroencephalogram (EEG) signals. Filtering process by means of band-pass finite impulse response (FIR) filter with the frequency range of 0.5–40 Hz is implemented at the outset to e...
Guardado en:
Autores principales: | , , , |
---|---|
Formato: | article |
Lenguaje: | EN |
Publicado: |
Elsevier
2021
|
Materias: | |
Acceso en línea: | https://doaj.org/article/bc0732238f8a492e90be0cd3e361b4bb |
Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
id |
oai:doaj.org-article:bc0732238f8a492e90be0cd3e361b4bb |
---|---|
record_format |
dspace |
spelling |
oai:doaj.org-article:bc0732238f8a492e90be0cd3e361b4bb2021-11-14T04:35:41ZEpileptic seizure identification in EEG signals using DWT, ANN and sequential window algorithm2666-222110.1016/j.socl.2021.100026https://doaj.org/article/bc0732238f8a492e90be0cd3e361b4bb2021-12-01T00:00:00Zhttp://www.sciencedirect.com/science/article/pii/S2666222121000150https://doaj.org/toc/2666-2221A patient-specific novel systematic methodology is described in this study for automatic seizure detection from raw electroencephalogram (EEG) signals. Filtering process by means of band-pass finite impulse response (FIR) filter with the frequency range of 0.5–40 Hz is implemented at the outset to eliminate different artifacts and noises mixed with raw EEG signals. As EEGs are highly non-linear and non-stationary signals in nature, discrete wavelet transform (DWT) is then used to analyze the signals in time-frequency domain. DWT with four level decomposition is performed using db6 mother wavelet for feature extraction. A new feature set, composed of eleven non-linear statistical features extracted from each sub-bands resulting from due to wavelet decomposition, is then fed to the input of artificial neural network (ANN) to classify the signal accurately. Finally, a novel algorithm named sequential window algorithm is carried out to improve the classification performance. 99.44% mean classification accuracy, 80.66% average sensitivity, 4.12 s mean latency and 0.2% average false positive rate (FPR) are achieved in this study. This study successfully reduces the latency time with more accuracy and significantly low FPR.Ramendra Nath BairagiMd ManiruzzamanSuriya PervinAlok SarkerElsevierarticleArtificial neural networkDiscrete wavelet transformEEG signal classificationFalse positive rateSeizure detectionSequential window algorithmInformation technologyT58.5-58.64Electronic computers. Computer scienceQA75.5-76.95ENSoft Computing Letters, Vol 3, Iss , Pp 100026- (2021) |
institution |
DOAJ |
collection |
DOAJ |
language |
EN |
topic |
Artificial neural network Discrete wavelet transform EEG signal classification False positive rate Seizure detection Sequential window algorithm Information technology T58.5-58.64 Electronic computers. Computer science QA75.5-76.95 |
spellingShingle |
Artificial neural network Discrete wavelet transform EEG signal classification False positive rate Seizure detection Sequential window algorithm Information technology T58.5-58.64 Electronic computers. Computer science QA75.5-76.95 Ramendra Nath Bairagi Md Maniruzzaman Suriya Pervin Alok Sarker Epileptic seizure identification in EEG signals using DWT, ANN and sequential window algorithm |
description |
A patient-specific novel systematic methodology is described in this study for automatic seizure detection from raw electroencephalogram (EEG) signals. Filtering process by means of band-pass finite impulse response (FIR) filter with the frequency range of 0.5–40 Hz is implemented at the outset to eliminate different artifacts and noises mixed with raw EEG signals. As EEGs are highly non-linear and non-stationary signals in nature, discrete wavelet transform (DWT) is then used to analyze the signals in time-frequency domain. DWT with four level decomposition is performed using db6 mother wavelet for feature extraction. A new feature set, composed of eleven non-linear statistical features extracted from each sub-bands resulting from due to wavelet decomposition, is then fed to the input of artificial neural network (ANN) to classify the signal accurately. Finally, a novel algorithm named sequential window algorithm is carried out to improve the classification performance. 99.44% mean classification accuracy, 80.66% average sensitivity, 4.12 s mean latency and 0.2% average false positive rate (FPR) are achieved in this study. This study successfully reduces the latency time with more accuracy and significantly low FPR. |
format |
article |
author |
Ramendra Nath Bairagi Md Maniruzzaman Suriya Pervin Alok Sarker |
author_facet |
Ramendra Nath Bairagi Md Maniruzzaman Suriya Pervin Alok Sarker |
author_sort |
Ramendra Nath Bairagi |
title |
Epileptic seizure identification in EEG signals using DWT, ANN and sequential window algorithm |
title_short |
Epileptic seizure identification in EEG signals using DWT, ANN and sequential window algorithm |
title_full |
Epileptic seizure identification in EEG signals using DWT, ANN and sequential window algorithm |
title_fullStr |
Epileptic seizure identification in EEG signals using DWT, ANN and sequential window algorithm |
title_full_unstemmed |
Epileptic seizure identification in EEG signals using DWT, ANN and sequential window algorithm |
title_sort |
epileptic seizure identification in eeg signals using dwt, ann and sequential window algorithm |
publisher |
Elsevier |
publishDate |
2021 |
url |
https://doaj.org/article/bc0732238f8a492e90be0cd3e361b4bb |
work_keys_str_mv |
AT ramendranathbairagi epilepticseizureidentificationineegsignalsusingdwtannandsequentialwindowalgorithm AT mdmaniruzzaman epilepticseizureidentificationineegsignalsusingdwtannandsequentialwindowalgorithm AT suriyapervin epilepticseizureidentificationineegsignalsusingdwtannandsequentialwindowalgorithm AT aloksarker epilepticseizureidentificationineegsignalsusingdwtannandsequentialwindowalgorithm |
_version_ |
1718429920507461632 |