Loading...

“Detection and Analysis of Spindle and K-complex Patterns and SWS in Sleep EEG Signals”

Najafi, Mahshid | 2011

721 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 41601 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Shamsollahi, Mohammad Bagher; Molaee-Ardekani, Behnam
  7. Abstract:
  8. According to necessity of analysis and detection of K-complexes and Sleep Spindles patterns which are the hallmarks of the second stage of sleep, in this thesis we aimed to introduce new methods in analysis and detection of aforementioned patterns in order to improve the results of previous methods. Also, we tried to find the relation between slow oscillations and spindles activity. In this project, in order to analysis the frequency components of Sleep Spindle, Bump modeling and STFT were used. Both of these methods confirm the spindles’ 8 Hz to 15 Hz frequency band and also their time duration between 0.5-2 seconds. On the other hand, we used modified matched filtering and also bump modeling along with Neural Networks and also ECOC classifier in order to automatically detect the spindles in sleep EEG. In the best case, the results of classifying bump modeling features with ECOC had a sensitivity of 99.4% when its E.D.R was only 14.14% which shows higher sensitivity along with lower E.D.R simultaneously.For analyzing the K-complexes, the bump modeling were used, and also for detecting this pattern in sleep EEG, Hilbert-Huang Transform and Bump modeling features, along with neural networks and ECOC were used. Therefore, we could introduce both of these methods with higher sensitivity and lower E.D.R than previous methods in detection of k-complexes.In addition, the power spectrum density of SWS and spindles activity and also amplitude and phase modulation between SWS and spindles activity were investigated. The results of these analyses show that the 8 Hz to 15 Hz frequency band in second stage of sleep EEG is mostly because of sleep spindles activity, while in other stages, this frequency band is related to other patterns and activities. The phase and amplitude modulation shows that between slow delta frequency band (1-2 Hz) and slow spindles activity (8-12 Hz), there is a strong modulation.
  9. Keywords:
  10. Detection ; Electroencephalography ; Sleep Spindles Pattern ; K-Complex Pattern ; Slow Oscillations ; Bump Modeling

 Digital Object List

 Bookmark

No TOC