Seizure Detection of EEG Signals Based on Multi-Channel Long and Short-Term Memory-Like Spiking Neural Model

International Journal of Neural Systems, Journal Year: 2024, Volume and Issue: 34(10)

Published: June 21, 2024

Seizure is a common neurological disorder that usually manifests itself in recurring seizure, and these seizures can have serious impact on person's life health. Therefore, early detection diagnosis of seizure crucial. In order to improve the efficiency this paper proposes new method, which based discrete wavelet transform (DWT) multi-channel long- short-term memory-like spiking neural P (LSTM-SNP) model. First, signal decomposed into 5 levels by using DWT obtain features components at different frequencies, series time-frequency coefficients are extracted. Then, used train LSTM-SNP model perform detection. The proposed method achieves high accuracy CHB-MIT dataset: 98.25% accuracy, 98.22% specificity 97.59% sensitivity. This indicates epilepsy show competitive performance.

Language: Английский

---

Multi-view brain network classification based on Adaptive Graph Isomorphic Information Bottleneck Mamba

Expert Systems with Applications, Journal Year: 2024, Volume and Issue: 267, P. 126170 - 126170

Published: Dec. 21, 2024

Language: Английский

---

Patient-independent seizure detection based on long-term iEEG and a novel lightweight CNN

Journal of Neural Engineering, Journal Year: 2023, Volume and Issue: 20(1), P. 016037 - 016037

Published: Jan. 10, 2023

Abstract Objective. Patient-dependent seizure detection based on intracranial electroencephalography (iEEG) has made significant progress. However, due to the difference in locations and number of iEEG electrodes used for each patient, patient-independent not been carried out. Additionally, current algorithms deep learning have outperformed traditional machine many performance metrics. they still shortcomings large memory footprints slow inference speed. Approach. To solve above problems study, we propose a novel lightweight convolutional neural network model combining Convolutional Block Attention Module (CBAM). Its is evaluated two long-term continuous datasets: SWEC-ETHZ TJU-HH. Finally, reproduce four other methods compare with our method calculate speed all methods. Main results. Our achieves 83.81% sensitivity (SEN) 85.4% specificity (SPE) dataset 86.63% SEN 92.21% SPE TJU-HH dataset. In particular, it takes only 11 ms infer 10 min (128 channels), its footprint 22 kB. Compared baseline methods, better but also smaller faster Significance. knowledge, this first iEEG-based study. This facilitates application future clinic.

Language: Английский

---

B2-ViT Net: Broad Vision Transformer Network With Broad Attention for Seizure Prediction

IEEE Transactions on Neural Systems and Rehabilitation Engineering, Journal Year: 2023, Volume and Issue: 32, P. 178 - 188

Published: Dec. 25, 2023

Seizure prediction are necessary for epileptic patients. The global spatial interactions among channels, and long-range temporal dependencies play a crucial role in seizure onset prediction. In addition, it is to search features vast space learn new generalized feature representations. Many previous deep learning algorithms have achieved some results automatic However, most of them do not consider channels together, only the representation space. To tackle these issues, this study, an novel bi-level programming model, B2-ViT Net, proposed spatio-temporal correlation features, which can characterize spatial, required model comprehensively due its strong broad capabilities. Sufficient experiments conducted on two public datasets, CHB-MIT Kaggle datasets. Compared with other existing methods, our has shown promising tasks, provides certain degree interpretability.

Language: Английский

---

Automatic Seizure Detection Using Multi-Input Deep Feature Learning Networks for EEG Signals

Journal of Sensors, Journal Year: 2024, Volume and Issue: 2024, P. 1 - 15

Published: Feb. 5, 2024

Epilepsy, a neurological disease associated with seizures, affects the normal behavior of human beings. The unpredictability epileptic seizures has caused great obstacles to treatment disease. automatic seizure detection method based on electroencephalogram (EEG) can assist experts in predicting improve efficiency. Epileptic cannot be achieved accurately using single-view characteristics signals. Moreover, manual feature extraction is time-consuming task. To design high-performance identification method, learning multi-view features becomes an indispensable part for detection. Therefore, paper proposes multi-input deep networks (MDFLN) model, which comprehensively considers from time domain and time–frequency (TF) EEG MDFLN model automatically extracts information signals through networks. Then, bidirectional long short-term memory (BLSTM) network used distinguish nonseizure events. Furthermore, effectiveness proposed structure verified two public datasets. experimental results demonstrate that classification accuracy at least 2.2% higher than features. achieves better performance CHB-MIT Bonn datasets 98.09% 98.4%, respectively. fine-tuned validation set also improves performance. Compare state-of-the-art methods, superior competence high sensitivity dataset. reduce consumption effectively clinical diagnosis treatment.

Language: Английский

---

On the Intersection of Signal Processing and Machine Learning: A Use Case-Driven Analysis Approach

Published: March 30, 2024

Recent advancements in sensing, measurement, and computing technologies have significantly expanded the potential for signal-based applications, leveraging synergy between signal processing Machine Learning (ML) to improve both performance reliability. This fusion represents a critical point evolution of systems, highlighting need bridge existing knowledge gap these two interdisciplinary fields. Despite many attempts literature this gap, most are limited specific applications focus mainly on feature extraction, often assuming extensive prior processing. assumption creates significant obstacle wide range readers. To address challenges, paper takes an integrated article approach. It begins with detailed tutorial fundamentals processing, providing reader necessary background knowledge. Following this, it explores key stages standard processing-based ML pipeline, offering in-depth review extraction techniques, their inherent solutions. Differing from literature, work offers application-independent introduces novel classification taxonomy techniques. Furthermore, aims at linking theoretical concepts practical demonstrates through use cases: spectral-based method condition monitoring rolling bearings wavelet energy analysis epilepsy detection using EEG signals. In addition contributions, promotes collaborative research culture by public repository relevant Python MATLAB codes. effort is intended support efforts ensure reproducibility results presented.

Language: Английский

---

Multiscale distribution entropy analysis of short epileptic EEG signals

Mathematical Biosciences & Engineering, Journal Year: 2024, Volume and Issue: 21(4), P. 5556 - 5576

Published: Jan. 1, 2024

<abstract> <p>This paper proposes an information-theoretic measure for discriminating epileptic patterns in short-term electroencephalogram (EEG) recordings. Considering nonlinearity and nonstationarity EEG signals, quantifying complexity has been preferred. To decipher abnormal EEGs, i.e., ictal interictal via recordings, a distribution entropy (DE) is used, motivated by its robustness on the signal length. In addition, to reflect dynamic inherent multiscale analysis incorporated. Here, two (MDE) methods using coarse-graining moving-average procedures are presented. Using popular datasets, Bonn Bern-Barcelona performance of proposed MDEs verified. Experimental results show that robust length thus reflecting over multiple time scales. consistent irrespective selection EEGs from entire recording. By evaluating Man-Whitney U test classification performance, can better discriminate than existing methods. Moreover, MDE with procedure performs marginally one coarse-graining. The experimental suggest applicable practical seizure detection applications.</p> </abstract>

Language: Английский

---

Seizure detection via deterministic learning feature extraction

Pattern Recognition, Journal Year: 2024, Volume and Issue: 153, P. 110466 - 110466

Published: April 10, 2024

Language: Английский

---

Automatic epileptic seizure detection based on persistent homology

Frontiers in Physiology, Journal Year: 2023, Volume and Issue: 14

Published: Dec. 12, 2023

Epilepsy is a prevalent brain disease, which quite difficult-to-treat or cure. This study developed novel automatic seizure detection method based on the persistent homology method. In this study, Vietoris–Rips (VR) complex filtration model was constructed EEG data. And applied to calculate VR barcodes describe topological changes of recordings. Afterward, as characteristics signals were fed into GoogLeNet for classification. The applicable multi-channel data analysis, where global information calculated and features are extracted by considering whole, without multiple calculations post-stitching. Three databases used evaluate proposed approach results showed that had high performances in epilepsy detection. obtained from CHB-MIT Database recordings revealed can achieve segment-based averaged accuracy, sensitivity specificity values 97.05%, 96.71% 97.38%, an event-based value 100% with 1.22 s average latency. addition, Siena Scalp Database, yields 96.42%, 95.23% 97.6%. Multiple tasks Bonn also achieved accuracy 99.55%, 98.63%, 98.28% 97.68%, respectively. experimental these three illustrate efficiency robustness our epileptic seizure.

Language: Английский

---

A novel end-to-end approach for epileptic seizure classification from scalp EEG data using deep learning technique

International Journal of Information Technology, Journal Year: 2023, Volume and Issue: 15(8), P. 4223 - 4231

Published: Sept. 22, 2023

Language: Английский

---