Using Explainable Artificial Intelligence to Obtain Efficient Seizure-Detection Models Based on Electroencephalography Signals

Sensors, Journal Year: 2023, Volume and Issue: 23(24), P. 9871 - 9871

Published: Dec. 16, 2023

Epilepsy is a condition that affects 50 million individuals globally, significantly impacting their quality of life. Epileptic seizures, transient occurrence, are characterized by spectrum manifestations, including alterations in motor function and consciousness. These events impose restrictions on the daily lives those affected, frequently resulting social isolation psychological distress. In response, numerous efforts have been directed towards detection prevention epileptic seizures through EEG signal analysis, employing machine learning deep methodologies. This study presents methodology reduces number features channels required simpler classifiers, leveraging Explainable Artificial Intelligence (XAI) for seizures. The proposed approach achieves performance metrics exceeding 95% accuracy, precision, recall, F1-score utilizing merely six five temporal domain with time window 1 s. model demonstrates robust generalization across patient cohort included database, suggesting feature reduction models-without resorting to learning-is adequate seizure detection. research underscores potential substantial reductions attributes channels, advocating training models strategically selected electrodes, thereby supporting development effective mobile applications

Language: Английский

---

Multiband seizure type classification based on 3D convolution with attention mechanisms

Computers in Biology and Medicine, Journal Year: 2023, Volume and Issue: 166, P. 107517 - 107517

Published: Sept. 25, 2023

Language: Английский

---

Detection of Anxiety-Based Epileptic Seizures in EEG Signals Using Fuzzy Features and Parrot Optimization-Tuned LSTM

Brain Sciences, Journal Year: 2024, Volume and Issue: 14(8), P. 848 - 848

Published: Aug. 22, 2024

In humans, epilepsy is diagnosed through electroencephalography (EEG) signals. Epileptic seizures (ESs) arise due to anxiety. The detection of anxiety-based challenging for radiologists, and there a limited availability EEG Data augmentation methods are required increase the number novel samples. An epileptic seizure arises anxiety, which manifests as variations in signal patterns consisting changes size shape signal. this study, anxiety signals were synthesized by applying data such random (RDA) existing from Bonn dataset. data-augmented processed using three algorithms-(i) fuzzy C-means-particle swarm optimization-long short-term memory (FCM-PS-LSTM), (ii) particle (PS-LSTM), (iii) parrot optimization LSTM (PO-LSTM)-for ESs via predicted accuracies detecting proposed algorithms-namely, (i) FCM-PS-LSTM, PS-LSTM, PO-LSTM-were about 98%, 98.5%, 96%, respectively.

Language: Английский

---

EEG-based Epileptic Seizure Detection using Deep Learning Techniques: A Survey

Neurocomputing, Journal Year: 2024, Volume and Issue: unknown, P. 128644 - 128644

Published: Sept. 1, 2024

Language: Английский

---

Detection of epileptic seizure events using pre‐trained convolutional neural network, VGGNet and ResNet

Expert Systems, Journal Year: 2023, Volume and Issue: unknown

Published: Sept. 12, 2023

Abstract Epilepsy is a life threatening neurological disorder. The person with epilepsy suffers from recurrent seizures. Sudden emission of electrical signal in the nerves human brain called seizure event. most widely used method for diagnosing analysing electroencephalogram signals short as EEG collected scalp patient. data are normally detection. If detected input dataset, then it can be considered presence Manual inspection laborious process. An automated system very crucial neurologists to identify In this paper, an detection presented using deep learning method, pre‐trained convolutional neural network architecture. Freely available dataset Temple University Hospital database study. CNN networks, VGGNet and ResNet classifying activities non‐seizure activities. CNNs extremely good features data. A large TUH provided multiple layers model. same fed models. results CNN, models assessed performance metrics accuracy, AUC, precision recall. All three gave compared state‐of‐the‐art works literature. comparison performed little higher giving 97% 96% 79%

Language: Английский

---

A hybrid EEG classification model using layered cascade deep learning architecture

Medical & Biological Engineering & Computing, Journal Year: 2024, Volume and Issue: 62(7), P. 2213 - 2229

Published: March 20, 2024

Language: Английский

---

Automated Multi-class Seizure-type Classification System using EEG Signals and Machine Learning Algorithms

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 136524 - 136541

Published: Jan. 1, 2024

Language: Английский

---

A Modified Transformer Network for Seizure Detection Using EEG Signals

International Journal of Neural Systems, Journal Year: 2024, Volume and Issue: unknown

Published: Oct. 25, 2024

Seizures have a serious impact on the physical function and daily life of epileptic patients. The automated detection seizures can assist clinicians in taking preventive measures for patients during diagnosis process. combination deep learning (DL) model with convolutional neural network (CNN) transformer effectively extract both local global features, resulting improved seizure performance. In this study, an enhanced named Inresformer is proposed detection, which combined Inception Residual extracting different scale features electroencephalography (EEG) signals to enrich feature representation. addition, replaces existing Feedforward layers two half-step enhance nonlinear representation model. architecture utilizes discrete wavelet transform (DWT) decompose original EEG signals, three sub-bands are selected signal reconstruction. Then, Co-MixUp method adopted solve problem data imbalance, processed sent information capture recognition. Finally, discriminant fusion performed results three-scale sub-signals achieve final achieves best accuracy 100% Bonn dataset average 98.03%, sensitivity 95.65%, specificity 98.57% long-term CHB-MIT dataset. Compared DL networks, holds significant potential clinical research applications competitive

Language: Английский

---

Epileptic Seizure Detection based on Hyperparameter Optimization using EEG Data

2022 International Conference on Sustainable Computing and Data Communication Systems (ICSCDS), Journal Year: 2023, Volume and Issue: unknown, P. 890 - 893

Published: March 23, 2023

Epilepsy is a common nerve disorder resulting in seizures. Seizure best defined as an abnormal electrical activity that can affect brain's function. EEG Signals are essential for epileptic seizure detection since they show the rise brain. Machine learning allows software applications to predict outcomes precisely without being explicitly programmed. Artificial Neural Network (ANN) popular classifier Learning (ML) categorizes data and detects seizures efficiently compromising performance. ANN architecture includes several parameters like neurons, activation function, optimizers so on. For accurate prediction, not all datasets need same set of these parameters. So tuning according dataset important. The proposed work focus on three different methods (Bayes Optimization, Gird Search Random Search) search values hyperparameters such batch size, epochs, rate, number optimizer support model achieve performance given dataset.

Language: Английский

---

Epileptic Seizure Detection Based on Feature Extraction and CNN-BiGRU Network with Attention Mechanism

Lecture notes in computer science, Journal Year: 2023, Volume and Issue: unknown, P. 308 - 319

Published: Jan. 1, 2023

Language: Английский

---