Feed

Electroencephalography (EEG) is a non-invasive neuroimaging technique that records electrical activity in the brain using electrodes placed on the scalp. It provides high temporal resolution, making it valuable for studying brain dynamics, cognitive processes, and neurological disorders. EEG is widely used in clinical and research settings for diagnosing epilepsy, sleep disorders, and mental health conditions. However, EEG signals are often contaminated by artifacts from ocular, muscular, and environmental sources, complicating accurate analysis. Traditional artifact removal methods, such as Independent Component Analysis (ICA) and wavelet transform, require significant computational resources and manual tuning, limiting their effectiveness in real-time applications. To overcome these challenges, this paper presents a deep learning-based framework for automated EEG denoising and simultaneous artifact removal while ensuring efficiency in real-time deployability. Various performance metrics such as relative-root-mean-square (RRMSE), structural similarity index measure (SSIM), and correlation, (CC) are measured to evaluate the model performance. The model achieves an average temporal and spectral RRMSE of 0.214 and 0.217 respectively, average SSIM of 0.964 and CC 0.963 of across various datasets. The model outperforms as compared to the state-of-the-art method with 295 × lesser parameters as compared to prior models. The model is able to denoise the EEG signals from various artifacts simultaneously. The proposed model demonstrates the potential for real-time deployment. The source code is available at https://github.com/dipayandewan94/EDGeNet.