SMU Data Science Review
Abstract
Convolutional neural networks (CNNs) have helped medical practitioners re-design the diagnosis of electroencephalogram (EEG) recordings, resulting in a level of accuracy and efficiency that matches or exceeds experts in the field. This study aims to explore the application of VGG19, an algorithm based on CNN architecture, to diagnose small subsets of neurodegenerative diseases, specifically amyotrophic lateral sclerosis (ALS), a disease that reduces the nervous system’s ability to control the muscles of the body. Since EEG recordings are scarce for ALS, this study will use a larger collection of EEG recordings taken on patients with more common neurological diseases, including but not limited to: Alzheimer’s, seizures, and epilepsy. This learning was then transferred over to a smaller dataset containing recordings of ALS patients. The model was able to achieve an accuracy of 80% on the large collection and 68% on the ALS dataset. These results demonstrate that it is possible to use the transfer learning from VGG19 and achieve accurate results when applying it to smaller sets of EEG recordings related to different types of neurological diseases. The current methods used to diagnose ALS are either invasive or expensive. With the increased ability to diagnose using EEG recordings, early detection of ALS and better patient care can also be achieved.
Recommended Citation
Lohia, Nibhrat; Mathew, Chris; and Shankel, Garrett
(2024)
"Applying Transfer Learning and Existing EEG Datasets to Identify Patients With ALS,"
SMU Data Science Review: Vol. 8:
No.
2, Article 4.
Available at:
https://scholar.smu.edu/datasciencereview/vol8/iss2/4
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License
