Related Experiment Videos
Cardiac artifact subspace identification and elimination in cognitive MEG data using time-delayed decorrelation
Tilmann H Sander1, Gerd Wübbeler, Andreas Lueschow
1Physikalisch-Technische Bundesanstalt, Laboratory 8.21, Universitätsklinikum Benjamin Franklin, Berlin, Germany. tsander@zedat.fu-berlin.de
IEEE Transactions on Bio-Medical Engineering
|April 11, 2002
Summary
Independent component analysis effectively identifies cardiac artifacts in magnetoencephalography (MEG) data. Suppressing these artifacts improves signal quality, reducing data epochs while preserving visually evoked signals.
Area of Science:
- Neuroscience
- Biophysics
- Signal Processing
Background:
- Physiological artifacts, such as cardiac signals, contaminate magnetoencephalographic (MEG) and electroencephalographic (EEG) recordings.
- Traditional artifact reduction methods include principal component analysis, signal-space projection, and regression, each with limitations.
Purpose of the Study:
- To evaluate the effectiveness of the time-delayed decorrelation algorithm for identifying and suppressing cardiac artifacts in MEG data.
- To introduce novel measures for assessing signal preservation after artifact suppression.
- To improve the signal quality of visually evoked fields in MEG recordings.
Main Methods:
- Application of the time-delayed decorrelation algorithm to raw MEG data from a visual stimulation experiment.
- Calculation of average cardiac activity to analyze field distributions.
- Suppression of the multidimensional cardiac artifact subspace.
- Introduction of geometrical and temporal measures to assess evoked signal preservation.
Main Results:
- Independent component analysis identified several components attributable to cardiac artifacts.
- Different cardiac excitation states produced similar field distributions and spectral properties, leading to combined artifact components.
- Suppression of cardiac and alpha wave artifacts reduced the number of epochs by half.
- Visually evoked signals were preserved after artifact suppression.
Conclusions:
- The time-delayed decorrelation algorithm can identify cardiac artifacts in MEG data, but a one-to-one assignment with physiological sources is not always justified.
- Multidimensional cardiac artifact subspace suppression effectively improves signal quality in MEG.
- Artifact reduction allows for efficient data processing by reducing the number of required epochs while maintaining signal integrity.