Enhanced Feature Extraction Method for EEG Based on Empirical Mode Decomposition

The traditional way of epileptic seizures detecting is through Electroencephalography (EEG) signal. As research on epilepsy prediction deepens, various issues such as muscle artifacts, eye blinking, and the inherent complexity of EEG signals have been addressed to varying degrees. However, in practical applications, there is a higher demand for detection performance. Therefore, this paper proposes an improved method based on empirical mode decomposition (EMD), which functions in the preprocessing of EEG signals to enhance relevant features and improve prediction results with broad transplantability. Firstly, the brain electrical signals are decomposed by EMD into a fixed number of Intrinsic Mode Functions (IMFs); then, the IMFs are optimized and recombined to enhance the feature according to the principle of orthogonal experiment design; finally, a Convolutional Neural Network (CNN) is employed for feature extraction and epilepsy classification prediction. The proposed method is evaluated using the EEG data corpus of the Temple University Hospital, and the results demonstrate a higher detection accuracy compared to other methods. This indicates that the method of inherent modal enhancement of features based on orthogonal experiment design is an effective preprocessing method for epilepsy detection.