Date of Award
2023
Degree Type
Thesis
Degree Name
Master of Science in Electrical Engineering (MSEE)
Specialization
Biomedical Engineering
Department
Electrical, Computer, and Biomedical Engineering
First Advisor
Kaushallya Adhikari
Abstract
Electroencephalogram (EEG) is a commonly used non-invasive method that acquires voltage of the brain during neural activities by using conventional disc electrodes. However, there are some fundamental limitations associated with EEG such as poor spatial resolution and signal-to-noise ratio (SNR), inaccurate data due to overlapping information, and artifact contamination. Also, the EEG software filtering solutions had the potential of distorting EEG data making the results unreliable. The tripolar electroencephalogram (tEEG) records brain signals by using the tripolar concentric ring electrodes (TCRE). Based on the surface Laplacian algorithm, signals recorded from TCRE overcome the limitations of EEG. tEEG has better SNR and spatial resolution, the ability of acquiring high frequency activities (up to 425 Hz so far) and can also perform muscle artifact rejection (electromyography (EMG) rejection). This study aimed to determine if the tEEG gives better performance in language mapping with the “language dominant hemisphere classification” and motor cortex mapping with “finger movements classification”. Our analysis shows that tEEG yields higher average classification accuracy when compared to EEG using spectrum data even with simple signal processing procedures.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Pham, Thao Nguyen Ngoc, "CONVOLUTIONAL NEURAL NETWORKS FOR CLASSIFICATION TEEG AND EEG" (2023). Open Access Master's Theses. Paper 2340.
https://digitalcommons.uri.edu/theses/2340