Eeg sinyallerinden disfaji hastalığının karakteristiklerinin belirlenmesi ve analizi

Dysphagia is a swallowing disorder that is usually associated with neurological diseases and negatively affects the quality of life, especially in elderly individuals. This study investigates the neurophysiological analysis of swallowing and motor imagery processes using EEG data and how this data can be used in dysphagia rehabilitation. Different experimental paradigms, such as natural swallowing, induced saliva swallowing, induced water swallowing, and induced tongue protrusion, were used in the experiments conducted on 30 right-handed individuals. Techniques such as Independent Component Analysis (ICA), Empirical Mode Decomposition (EMD), band-pass filtering, and Common Spatial Pattern (CSP) analysis were applied in the preprocessing of the data. These preprocessing methods provided a more accurate analysis by reducing the noise in the EEG data. The differences between the resting and imagery stages were clearly separated in the classification tasks performed with traditional machine learning techniques and deep learning methods. In addition to ensemble-based algorithms such as Random Forest, AdaBoost, and Bagging, Convolutional Neural Networks (CNN) from deep learning methods were also applied. In addition, the multi-scale spatial attention network (MS-SAN) model distinguished neurophysiological differences between motor imagery and resting states, especially in delta and theta frequency bands, with high accuracy. The results show that detecting motor imagery and resting stages with EEG data has great potential in dysphagia treatment and motor rehabilitation applications. This study highlights the potential of EEG-based brain-computer interface (BCI) technologies, machine learning, and deep learning methods in dysphagia rehabilitation. It reveals the importance of research in this area for clinical applications. Keywords: Electroencephalography, Machine Learning, Deep Learning, BCI, Swallowing