Abstract:
his study explores the use of deep learning methods, particularly Deep Convolutional Neural Networks (Deep CNN) and Cycle Generative Adversarial Networks (Cycle GAN), to purify neurophysiological signals such as EEG and PPG from unwanted artifacts. EEG signals are often contaminated by EOG and EMG artifacts, while PPG signals suffer from motion artifacts and baseline drifts. Our models have shown a significant improvement in signal quality compared to traditional techniques, highlighting the potential of deep learning architectures to enhance the processing of neurophysiological signals and biomedical applications.
