EEGWaveNet: Multiscale CNN-Based Spatiotemporal Feature Extraction for EEG Seizure Detection

Thuwajit, Punnawish; Rangpong, Phurin; Sawangjai, Phattarapong; Autthasan, Phairot; Chaisaen, Rattanaphon; Banluesombatkul, Nannapas; Boonchit, Puttaranun; Tatsaringkansakul, Nattasate; Sudhawiyangkul, Thapanun; Wilaiprasitporn, Theerawit

doi:10.1109/tii.2021.3133307

Cited by 76 publications

(33 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Deep learning baselines are EEGNet, EnK-EEGNet, EEG-WaveNet [49], and CE-stSENet. In addition, we improved EnK-EEGNet by replacing its summation representation matrix R with the average representation matrix in (7), named as Avg-EnK-EEGNet.…”

Section: Experimental Results In Seizure Subtype Classificationmentioning

confidence: 99%

TIE-EEGNet: Temporal Information Enhanced EEGNet for Seizure Subtype Classification

Peng

Zhao

Jiang

et al. 2022

IEEE Trans. Neural Syst. Rehabil. Eng.

View full text Add to dashboard Cite

show abstract

Section: Experimental Results In Seizure Subtype Classificationmentioning

confidence: 99%

TIE-EEGNet: Temporal Information Enhanced EEGNet for Seizure Subtype Classification

Peng

Zhao

Jiang

et al. 2022

IEEE Trans. Neural Syst. Rehabil. Eng.

View full text Add to dashboard Cite

show abstract

“…A comparison with the DLEK-GP [ 35 ] method adopts the classic common spatial pattern (CSP) and discriminative log-Euclidean kernel-based Gaussian process for distinguishing epileptic EEG signals. A further comparison with the EEGWaveNet [ 36 ] approach utilizes trainable depth-wise convolutions as discriminative filters to simultaneously gather features from each EEG channel and separate the signal into multiscale resolution. Lastly, the proposed CE-stSENet [ 37 ] proposes channel-embedding spectral-temporal squeeze-and-excitation network which can capture hierarchical multidomain representations in a unified manner with a variant of squeeze-and-excitation block.…”

Section: Experiments and Resultsmentioning

confidence: 99%

“…For subject-independent classification, also use EEGWaveNet [ 36 ] as one of the base line methods. A representative method with the sparse representation-based epileptic seizure classification based on the dictionary learning with the homotopy (DLWH) algorithm is proposed [ 38 ].…”

Section: Experiments and Resultsmentioning

confidence: 99%

Synthetic Epileptic Brain Activities with TripleGAN

Jie

Zhou

et al. 2022

Computational and Mathematical Methods in Medicine

View full text Add to dashboard Cite

Epilepsy is a chronic noninfectious disease caused by sudden abnormal discharge of brain neurons, which leads to intermittent brain dysfunction. It is also one of the most common neurological diseases in the world. The automatic detection of epilepsy based on electroencephalogram through machine learning, correlation analysis, and temporal-frequency analysis plays an important role in epilepsy early warning and automatic recognition. In this study, we propose a method to realize EEG epilepsy recognition by means of triple genetic antagonism network (GAN). TripleGAN is used for EEG temporal domain, frequency domain, and temporal-frequency domain, respectively. The experiment was conducted through CHB-MIT datasets, which operated at the latest level in the same industry in the world. In the CHB-MIT dataset, the classification accuracy, sensitivity, and specificity exceeded 1.19%, 1.36%, and 0.27%, respectively. The crossobject ratio exceeded 0.53%, 2.2%, and 0.37%, respectively. It shows that the established deep learning model of TripleGAN has a good effect on EEG epilepsy classification through simulation and classification optimization of real signals.

show abstract

“…Complementary to CNN, residual blocks have been added that encode temporal information and prevent performance degradation [14]. Promising examples include heart rate estimation [15], blood glucose monitoring [16], and seizure detection [17].…”

Section: Related Workmentioning

confidence: 99%

RRWaveNet: A Compact End-to-End Multi-Scale Residual CNN for Robust PPG Respiratory Rate Estimation

Osathitporn¹,

Sawadwuthikul²,

Thuwajit³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Respiratory rate (RR) is an important biomarker as RR changes can reflect severe medical events such as heart disease, lung disease, and sleep disorders. Unfortunately, however, standard manual RR counting is prone to human error and cannot be performed continuously. This study proposes a method for continuously estimating RR, RRWaveNet. The method is a compact end-to-end deep learning model which does not require feature engineering and can use low-cost raw photoplethysmography (PPG) as input signal. RRWaveNet was tested subject-independently and compared to baseline in three datasets (BIDMC, CapnoBase, and WESAD) and using three window sizes (16, 32, and 64 seconds). RRWaveNet outperformed current state-ofthe-art methods with mean absolute errors at optimal window size of 1.66 ± 1.01, 1.59 ± 1.08, and 1.92 ± 0.96 breaths per minute for each dataset. In remote monitoring settings, such as in the WESAD dataset, we apply transfer learning to two other ICU datasets, reducing the MAE to 1.52 ± 0.50 breaths per minute, showing this model allows accurate and practical estimation of RR on affordable and wearable devices. Our study shows feasibility of remote RR monitoring in the context of telemedicine and at home.

show abstract

EEGWaveNet: Multiscale CNN-Based Spatiotemporal Feature Extraction for EEG Seizure Detection

Cited by 76 publications

References 32 publications

TIE-EEGNet: Temporal Information Enhanced EEGNet for Seizure Subtype Classification

TIE-EEGNet: Temporal Information Enhanced EEGNet for Seizure Subtype Classification

Synthetic Epileptic Brain Activities with TripleGAN

RRWaveNet: A Compact End-to-End Multi-Scale Residual CNN for Robust PPG Respiratory Rate Estimation

Contact Info

Product

Resources

About