|
Activity Number:
|
359
|
|
Type:
|
Invited
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #307777 |
|
Title:
|
Methods for Signal Extraction from EEG Time Series with Application to Large Studies of Sleep
|
|
Author(s):
|
Chongzhi Di*+ and Ciprian M. Crainiceanu
|
|
Companies:
|
Johns Hopkins University and Johns Hopkins University
|
|
Address:
|
Department of Biostatistics, Baltimore, MD, 21205,
|
|
Keywords:
|
|
|
Abstract:
|
Sleep related indices are believed to be causally associated with several diseases, such as: hypertension, sleepiness and cardiovascular disease. A fundamental measure of sleep is obtained by monitoring electric brain signals (EEG) from one or more channels during sleep. EEG time series are then processed in short time windows (5 or 30 seconds) using Fast Fourier Transform (FFT) and filtered using four scientifically relevant bands: alpha, beta, delta, and theta. The fraction of power in each band is estimated in each time window. These estimates are noisy versions of the underlying power spectrum where the noise is a mixture of filtering resistant noise in the original EEG signal measurement. Penalized spline smoothing for each power band of each subject is used to reduce noise. Signal extraction is performed using two complementary strategies.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2007 program |
