Loading...

High Frequency Oscillation Detection in Brain Electrical Signals Using Tensor Decomposition

Yousefi Mashhoor, Reza | 2024

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 57672 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Hajipour, Sepideh
  7. Abstract:
  8. High-frequency oscillations (HFOs) in brain electrical signals are activities within the 80–500 Hz frequency range that are distinct from the baseline and include at least four oscillatory cycles. Research indicates that HFOs could serve as potential biomarkers for neurological disorders. Manual detection of HFOs is time-consuming and prone to human error, making automated HFO detection methods increasingly necessary. These automated methods typically rely on the signal's energy within the HFO frequency band. Tensor decompositions are mathematical models capable of extracting hidden information from multidimensional data. Due to the multidimensional nature of brain electrical signals, tensor decomposition is particularly suitable for analyzing such signals. This study investigates the use of tensor decompositions for effective feature extraction and dimensionality reduction to enable automated HFO detection in brain signals. The proposed method involves tensorizing the data using time-frequency analysis, clustering, filter extraction, feature extraction, and classification. Results demonstrate that simultaneously utilizing temporal and spectral information in filter extraction improves sensitivity by 6.59% and accuracy by 3.49% compared to using only temporal information. Furthermore, clustering with simultaneous temporal and spectral information enhances sensitivity by 29.4% compared to the non-clustering approach
  9. Keywords:
  10. Brain Signal ; High Frequency Oscillations (HFO) ; Tensor Decomposition ; Electrocorticography (ECOG) ; Epilepsy ; Clustering ; Feature Extraction ; Filters ; Seizure Detection

 Digital Object List

 Bookmark

No TOC