An Improved EEG Signal Feature Selection Paradigm for Migraine Detection

Recurrent headaches, autonomic nervous system dysfunction, nausea, and vomiting are the hallmarks of migraine, a complicated neurological illness that has a substantial negative influence on suffering individuals. With cutting-edge deep learning techniques and electroencephalography (EEG) data, this study presents a novel approach to automatic migraine detection. The Improved novel Hybrid Feature Extraction for Migraine Detection (IHFEMD) model that has been suggested takes a thorough strategy that includes pre-processing the data, extracting features, selecting features using a hybrid optimization technique, and utilizing a hybrid deep learning framework. To create a highly refined dataset, migraine sufferers' raw EEG data must first undergo a thorough pre-processing step that includes data cleaning and Min-Max normalization. As a result, relevant migraine-related data are obtained using time-frequency analysis methods such as Continuous Wavelet Transform (CWT) and Short-Time Fourier Transform (STFT). The Elephant Herding Optimization (EHO) method is integrated with the Bald Eagle Optimization (BEO) method to optimize the process. Due to this, a much more accurate selection is made. Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) are used in tandem by the detection framework to combine the advantages of both network types and accomplish migraine detection. Complicated EEG can be identified with the help of this combined approach. The model has been put into practice with the Matlab platform. The result shows an accuracy of 99% and is preferred as a potential diagnostic tool. The novelty of the proposed work focuses on integrating deep learning frameworks with complex pre-processing, sophisticated feature extraction, and hybrid optimization techniques.