Deep Learning in EEG-Based BCIs: A Comprehensive Review of Transformer Models, Advantages, Challenges, and Applications

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 127271 - 127301

Published: Jan. 1, 2023

Brain-computer interfaces (BCIs) have undergone significant advancements in recent years. The integration of deep learning techniques, specifically transformers, has shown promising development research and application domains. Transformers, which were originally designed for natural language processing, now made notable inroads into BCIs, offering a unique self-attention mechanism that adeptly handles the temporal dynamics brain signals. This comprehensive survey delves transformers providing readers with lucid understanding their foundational principles, inherent advantages, potential challenges, diverse applications. In addition to discussing benefits we also address limitations, such as computational overhead, interpretability concerns, data-intensive nature these models, well-rounded analysis. Furthermore, paper sheds light on myriad BCI applications benefited from incorporation transformers. These span motor imagery decoding, emotion recognition, sleep stage analysis novel ventures speech reconstruction. review serves holistic guide researchers practitioners, panoramic view transformative landscape. With inclusion examples references, will gain deeper topic its significance field.

Language: Английский

---

STEADYNet: Spatiotemporal EEG analysis for dementia detection using convolutional neural network

Cognitive Neurodynamics, Journal Year: 2024, Volume and Issue: 18(5), P. 3195 - 3208

Published: July 19, 2024

Language: Английский

---

Exploring Rhythms and Channels-Based EEG Biomarkers for Early Detection of Alzheimer's Disease

IEEE Transactions on Emerging Topics in Computational Intelligence, Journal Year: 2024, Volume and Issue: 8(2), P. 1609 - 1623

Published: Jan. 26, 2024

There is no treatment that permanently cures Alzheimer's disease (AD); however, early detection can alleviate the severe effects of disease. To support different stages AD (e.g., mild, moderate), key aim this study to develop a computer aided diagnostic (CAD) framework include long short-term memory (LSTM) network using massive multi-channel electroencephalogram (EEG) data. Although EEG rhythms and channels jointly possess important biomarkers may be used for diagnosis AD, but traditional methods did not explore issue in any research. address problem, introduces new identify optimal required AD. The proposed was tested on real-time dataset. results reveal together, gamma beta channels, Cz, F4, P4, T6, Pz were most reliable biomarker identifying LSTM based model yielded best performance. Additionally, another mild cognitive impairment (MCI) dataset test approach, excellent (accuracy>99%). will useful creating CAD system perform automatic diagnosis.

Language: Английский

---

Unlocking the Potential of EEG in Alzheimer's Disease Research: Current Status and Pathways to Precision Detection

Brain Research Bulletin, Journal Year: 2025, Volume and Issue: unknown, P. 111281 - 111281

Published: March 1, 2025

Alzheimer's disease (AD) affects millions of individuals worldwide and is considered a serious global health issue due to its gradual neuro-degenerative effects on cognitive abilities such as memory, thinking, behavior. There no cure for this but early detection along with supportive care plan may aid in improving the quality life patients. Automated AD challenging because symptoms vary patients genetic, environmental, or other co-existing conditions. In recent years, multiple researchers have proposed automated methods using MRI fMRI. These approaches are expensive, poor temporal resolution, do not offer real-time insights, proven be very accurate. contrast, only limited number studies explored potential Electroencephalogram (EEG) signals detection. present cost-effective, non-invasive, high-temporal-resolution alternative Despite their potential, application EEG research remains under-explored. This study reviews publicly available datasets, variety machine learning models developed detection, performance metrics achieved by these methods. It provides critical analysis existing approaches, highlights challenges, identifies key areas requiring further investigation. Key findings include detailed evaluation current methodologies, prevailing trends, gaps field. What sets work apart in-depth Disease providing stronger more reliable foundation understanding role area.

Language: Английский

---

Multimodal mixing convolutional neural network and transformer for Alzheimer’s disease recognition

Expert Systems with Applications, Journal Year: 2024, Volume and Issue: 259, P. 125321 - 125321

Published: Sept. 6, 2024

Language: Английский

---

Understanding the Role of Self-Attention in a Transformer Model for the Discrimination of SCD From MCI Using Resting-State EEG

IEEE Journal of Biomedical and Health Informatics, Journal Year: 2024, Volume and Issue: 28(6), P. 3422 - 3433

Published: April 18, 2024

The identification of EEG biomarkers to discriminate Subjective Cognitive Decline (SCD) from Mild Impairment (MCI) conditions is a complex task which requires great clinical effort and expertise. We exploit the self-attention component Transformer architecture obtain physiological explanations model's decisions in discrimination 56 SCD 45 MCI patients using resting-state EEG. Specifically, an interpretability workflow leveraging attention scores time-frequency analysis epochs through Continuous Wavelet Transform proposed. In classification framework, models are trained validated with 5-fold cross-validation evaluated on test set obtained by selecting 20% total subjects. Ablation studies hyperparameter tuning tests conducted identify optimal model configuration. Results show that best performing model, achieves acceptable results both epochs' patients' classification, capable finding specific patterns highlight changes brain activity between two conditions. demonstrate potential weights as tools guide experts understanding disease-relevant features could be discriminative MCI.

Language: Английский

---

Early Diagnosis of Alzheimer’s Disease in Human Participants Using EEGConformer and Attention-Based LSTM During the Short Question Task

Diagnostics, Journal Year: 2025, Volume and Issue: 15(4), P. 448 - 448

Published: Feb. 12, 2025

Background/Objectives: Alzheimer’s disease (AD) is a progressive neurodegenerative disorder advancing through subjective cognitive decline (SCD), mild impairment (MCI), and dementia, making early diagnosis crucial. Electroencephalography (EEG) non-invasive, cost-effective alternative to advanced neuroimaging for detecting neural changes. While most studies focus on resting-state EEG or handcrafted features with traditional machine learning, deep learning (DL) offers promising tool automated analysis. This study classified the AD spectrum (SCD, MCI, AD) using recorded during task-based conditions. Specifically, was simple yes/no question-answering task, mimicking everyday activities, explored. We hypothesized that brain activity tasks involving listening, comprehension, response execution provides diagnostic insights. Methods: collected 1 min of approximately 3 from 20, 28, 10 participants SCD, AD, respectively. Task data included accuracy reaction time. After minimal preprocessing, two DL models, attention long short-term memory EEGConformer, were used binary (e.g., SCD vs. MCI) three-class classification. Results: Task-based outperformed EEG, 5–15% improvement in accuracy. The area under curve (AUC) results consistently demonstrated superior classification performance compared across all group distinctions. No significant difference observed between models. Conclusions: proposed approach via offering greater by leveraging

Language: Английский

---

Investigating Brain Lobe Biomarkers to Enhance Dementia Detection Using EEG Data

Cognitive Computation, Journal Year: 2025, Volume and Issue: 17(2)

Published: April 1, 2025

Language: Английский

---

A new quantum-inspired pattern based on Goldner-Harary graph for automated alzheimer’s disease detection

Cognitive Neurodynamics, Journal Year: 2025, Volume and Issue: 19(1)

Published: May 10, 2025

Abstract Alzheimer's disease (AD) is a common cause of dementia. We aimed to develop computationally efficient yet accurate feature engineering model for AD detection based on electroencephalography (EEG) signal inputs. New method: retrospectively analyzed the EEG records 134 and 113 non-AD patients. To generate multilevel features, discrete wavelet transform was used decompose input EEG-signals. devised novel quantum-inspired EEG-signal extraction function 7-distinct different subgraphs Goldner-Harary pattern (GHPat), selectively assigned specific subgraph, using forward-forward distance-based fitness function, each block textural extraction. extracted statistical features standard moments, which we then merged with features. Other components were iterative neighborhood component analysis selection, shallow k-nearest neighbors, as well majority voting greedy algorithm additional voted prediction vectors select best overall results. With leave-one-subject-out cross-validation (LOSO CV), our attained 88.17% accuracy. Accuracy results stratified by channel lead placement brain regions suggested P4 parietal region be most impactful. Comparison existing methods: The proposed outperforms methods achieving higher accuracy approach, ensuring robustness generalizability. Cortex maps generated that allowed visual correlation channel-wise various regions, enhancing explainability.

Language: Английский

---

EEG Data Analysis Techniques for Precision Removal and Enhanced Alzheimer’s Diagnosis: Focusing on Fuzzy and Intuitionistic Fuzzy Logic Techniques

Signals, Journal Year: 2024, Volume and Issue: 5(2), P. 343 - 381

Published: May 31, 2024

Effective management of EEG artifacts is pivotal for accurate neurological diagnostics, particularly in detecting early stages Alzheimer’s disease. This review delves into the cutting-edge domain fuzzy logic techniques, emphasizing intuitionistic systems, which offer refined handling uncertainties inherent data. These methods not only enhance artifact identification and removal but also integrate seamlessly with other AI technologies to push boundaries analysis. By exploring a range approaches from standard protocols advanced machine learning models, this paper provides comprehensive overview current strategies emerging management. Notably, fusion neural network models illustrates significant advancements distinguishing between genuine activity noise. synthesis improves diagnostic accuracy enriches toolset available researchers clinicians alike, facilitating earlier more precise neurodegenerative diseases. The ultimately underscores transformative potential integrating diverse computational setting new analysis paving way future innovations medical diagnostics.

Language: Английский

---