Docstoc

Artifact cancellation Spectral components

Document Sample
Artifact cancellation Spectral components Powered By Docstoc
					Artifact cancellation and
nonparametric spectral
         analysis
                Outline
 Artifact processing
 Artifact cancellation
 Nonparametric spectral analysis
                    Introduction
 Artifact   processing
     Rejectioncancellation
     Rejection main alternative
       • one would hope to retain data
     Cancellation requirements
       • clinical information
       • no new artifacts
       • spike detectors
     Additive/multiplicative model
     Artifact reduction using linear filtering
          Artifact cancellation
 Using linearly combined reference signals
 Adaptive artfact cancellation using linearly
  combined reference signals
 Using filtered reference signals
        Linearly combined reference
                   signals
   Eye movements &
    blinks
       several referene
        signals
       positioning
       additive model
       EOG linearly
        trasferred to EEG
         • weights
                    In detail
 Uncorrelated
 Mean  square error
 Minimization, differentation
 Spatial correlation, cross correlation
     fixed over time
     zero gradient
 Estimation
     blinks, eye-movements at onset
                   In detail 2
 Number  of reference signals
 Only EOG cancelled
 ECG
 Rejection used a lot (in MEG)
     expect when lots of blinks (ssp)
              Adaptive version
 Time-varying  changes
 Tracking of slow changes
 Adaptive algorithm
     LMS
     weight(s) function of time
       • optimal solution changes with time
     method of steepest descent
     negative error gradient vector
                         In detail
   Parameter selection
       time
       noise
   Expectation
       instantaneous value
       zero setting
       performance
        estimation
       fluctuation of weights
      Filtered reference signals
 EOGpotentials exhibit frequency
 dependence
     in trasfer to EEG sensor through tissue
     blinks and eye movements
 Improved  cancellation with transfer
 function replacement
     spatial and temporal information
     v0 estimation
     FIR (lengths)
                          Details
   Stationary processes
       Second order characterisrics
       Correlation information fixed
                   Details 2
 No   a priori information
     can be implemented, modified error
 Also   adaptive version exists
     a priori impulse responses calculated at
      calibration
Nonparametric spectral analysis
 Richer  characterization of background
  activity that with 1D histograms
 EEG rhythms
 Correlate signals with sines and cosines
 When?
     Gaussian stationary signals
       • Stationary estimatation
     Normal spontaneous waking activity
              Nonparametric 2
 Fourier-based        power spectrum analysis
     no modeling assumptions
 Spectral    parameters
     interpretation
  Fourier-based power spectrum
             analysis
 Power spectrum characterized by
 correlation function (stationary)
     If ergodic, approximate with time average
      estimator (negative lags)
     combination called periodogram
     equals squared magnitude of DFT
         Fourier considerations
 Periodogram          biased
     window dependent (convolution)
     smearing (main lobe)
     leakage (side lobes)
       • synchronized rhythm better described by power in
         frequency band
     variance periogoram
       • does not approach zero with sample increase
             consistency
                  Periodogram
 Windowing       and averaging
     leakage & periodogram variance reduction
 Windows
     from rectangular to smaller sidelobes
       • wider main lobe, spectral resolution
 Variance     reduction
     nonoverlapping segments, averaging
       • resolution decrease, trade-off
       • combinations, degree of overlap
And then what...
              Spectral parametrs
 Resulting  power spectrum often not
 readilty interpreted
     Condensed into compact set of parameters
     feature extraction
       • parameters describing prominent features of the
         spectrum
             peaks, frequencies
       • general usage
             Spectral choices
 Visual   inspection
     format selection
     assessing represantiveness
 Scaling
     scope of the analysis
             Parameters
 Power  in frequency bands
 Peak frequency
 Spectral slope
 Hjort descriptors
 Spectral purity index
      Power in frequency bands
 Fixed/statistical   bands
     alpha, beta, theta etc.
     from data
 Ratio   of, absolute power
     comparison, nonphysiological factors
             Peak frequency
 Frequency, amplitude, width
 ad hoc methods for determining peaks
 more than just maximum
     median, mean
                 Spectral slope
 EEG     activity made of 2 component
     rhythmic, unstructured
      on decay of high frequency
 Based
 components
     one parameters approximation
       • least squares error
 Quantifcation of EEG
 Preconditioning of power estimate
                 Hjort descriptors
   Spectral moments
       H0 (activity)
       H1 (mobility)
       H2 (complexity)
 Signal power,
  dominant frequency,
  bandwidth
 Effectively in time
  domain
 Clinically useful
      Spectral purity index (SPI)
 Heuristic
 Reflectssignal bandwidth (H2)
 How well signal is described by a single
  frequency
     noise susceptibility
                          Summary

 Artifact   cancellation
     reference signals
     linear combinations, filtering
       • adaptive version(s)
 Spectral     parameters
     nonparametric
       • no modelling
     parametric
       • interpretation

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:6
posted:10/9/2011
language:
pages:27