Frequency-band based synthetic EEG generation
Master thesis
Permanent lenke
https://hdl.handle.net/11250/3162963Utgivelsesdato
2024Metadata
Vis full innførselSamlinger
Sammendrag
Electroencephalography (EEG) is a key technology in brain research because it offers non-invasive, high-temporal resolution information about the brain's electrical activity. This technology is now widely used in clinical medicine, psychology, and neuroscience due to its portability, affordability, and safety compared to alternatives. The capacity of EEG to capture oscillatory brain activity has contributed to a better understanding of cognitive and perceptual functions such as attention, memory, and perception. The collection of EEG data, particularly in realistic situations, presents significant challenges due to ethical considerations and subject availability, despite its widespread application. As a result, research and application in the field of Brain-Computer Interfaces (BCI) and other related fields are limited.
This thesis addresses a significant gap in the development of synthetic EEG data by exploring a novel approach to the generation of synthetic EEG data within specified frequency bands, specifically the alpha band (8-12 Hz) and the theta band (4-8 Hz). These bands are critical because they are linked to various cognitive states. They play a major role in the advancement of neuroscientific research and applications of neurotechnology. Unlike previous synthetic EEG data generation research, this study uses Generative Adversarial Networks (GAN) to synthesize frequency band-based EEG data that replicates the distinct properties of the frequency bands. This method offers new opportunities for the efficient and moral investigation of brain activity in addition to overcoming the present constraints on data availability. All in all, this work will contribute to filling a gap in the literature and lay the groundwork for future advancements in generating frequency band-based EEG data for medical and research purposes.