Classic EEG (ERPs)/ Advanced EEG

30 downloads 66 Views 3MB Size Report
http://members.arstechnica.com/x/albino_eatpod/specific-eeg-states.gif. • Gamma ... Evoked frequency. Adapted ... Sylvain Baillet's presentation at HBM 2008.
Classic EEG (ERPs)/ Advanced  EEG Quentin Noirhomme

Outline • • • •

Origins of MEEG Origins of MEEG Event‐related potentials Time‐frequency decomposition i f d ii Source reconstruction

Before to start Before to start • EEGlab • Fieldtrip (included in spm)

Part I: Origins Part I: Origins • EEG Discovered by Hans  Berger in 1924 • Non invasive measure of  electrical brain activityy

Origins: MEG Origins: MEG • 1968

Origins

Baillet et al., IEEE Sig. Proc. Mag., 2001

Origins: Potentials Origins: Potentials

Origins

Baillet et al., IEEE Sig. Proc. Mag., 2001

M/EEG vs. fMRI M/EEG vs. fMRI

Raw EEG Raw EEG

EEG in coma EEG in coma Burst Suppression

Isoelectric

Alpha coma

Fp2‐T4 Fp2 T4 T4‐02 Fp2‐C4 C4‐02 Fp1‐T3 T3‐01 T3 01 Fp1‐C3 50 V 50 µV

C3‐01

20 µV

50 µV 1 s

1 s

20 µV 1 s

Thömke et al. BMC Neurology 2005 5:14  doi:10.1186/1471‐2377‐5‐14

EEG in sleep EEG in sleep

http\\:www.benbest.com

EEG Rhythms EEG Rhythms

• Gamma : > 30 Hz http://members.arstechnica.com/x/albino_eatpod/specific‐eeg‐states.gif

EEG events EEG events Burst

Spikes

http://members.arstechnica.com/x/albino_eatpod/specific‐eeg‐states.gif

Part II: Event‐Related Part II: Event Related potentials potentials

Wolpaw et al., 2000

Averaging Adapted from Tallon‐Bau udry and Bertrrand, 1999

Average potential (across trials/ subjects) relative to some specific event in time

Preprocessing 1. 1 2. 3 3. 4. 5.

Filtering Segmentation Artifact rejection if j i Averaging Baseline removal

Filtering • Why filter? – EEG consists of a signal plus noise – Some of the noise is sufficiently different in frequency  content from the signal that it can be suppressed  simply by attenuating different frequencies, thus  i l b tt ti diff tf i th making the signal more visible • Non‐neural physiological activity (skin/sweat  potentials) • Noise from electrical outlets • Highpass filter to remove drift due to sweating, … • Notch filter to remove the line noise (50‐60Hz) • Low‐pass filter (often 30Hz for ERP)

Segmentation

Artifacts

Artifacts

http://www.bci2000.org

Artifacts

http://www.bci2000.org

Artifacts

http://www.bci2000.org

Artifacts

http://www.bci2000.org

Artifact rejection Artifact rejection Visual inspection of the data Visual inspection of the data Thresholding (e.g., everything above 100µV) S i i l Statistical method  h d Independent component analysis – good for  blinks and other visual artifacts • Help if you have EOG and EMG channels p y • Do not trust automatic methods • • • •

Averaging

Averaging • Assumes that only the EEG noise varies from trial to trial ssu es a o y e o se a es o a o a • But – amplitude and latency will vary

Averaging: effects of variance

Latency L t variation i ti can b be a significant problem

Averaging • Assumes that only the EEG noise varies from trial to trial ssu es a o y e o se a es o a o a • But – amplitude and latency will vary • S/N ratio increases as a function of the square root of the  number of trials.  • It’s always better to try to decrease sources of noise than  to increase the number of trials to increase the number of trials.

Baseline correction Baseline correction • Remove Remove the mean of the recorded baseline  the mean of the recorded baseline (e.g., ‐200 ms to 0 ms) • Variation in baseline duration can induce  Variation in baseline duration can induce change in potential amplitude • Individually for each electrode I di id ll f h l d • SPM does it automatically while segemting  the data

Part III: Time‐frequency Part III: Time frequency decomposition

Adapted from Tallon‐Baudry and Bertrand, 1999

Evoked frequency Evoked frequency

Adapted from Tallon‐Baudry and Bertrand, 1999

Induced frequency decomposition Induced frequency decomposition

Adapted from Tallon‐Baudry and Bertrand, 1999

Induced frequency decomposition Induced frequency decomposition

Adapted from Tallon‐Baudry and Bertrand, 1999

Time‐frequency Time frequency decomposition decomposition

Adapted from Tallon‐Baudry and Bertrand, 1999

Continuous Morlet wavelet Continuous Morlet wavelet

http://amouraux.webnode.com/.

Analysis • Grand mean  Grand mean ‐>> Average across subject Average across subject • Convert ERP or TF decomposition into  images – =>    first/second‐level analysis fi t/ dl l l i

• Source reconstruction  – =>  first/second‐level analysis

1st Level Analysis Level Analysis • select select periods or time points in peri periods or time points in peri‐stimulus stimulus time  time Choice made a priori.

• sum over all time points

Part IV: Source reconstruction Part IV: Source reconstruction

From www.imt.uni‐luebeck.de, 2008

Source reconstruction Source reconstruction 1. Forward Model 1 Forward Model 2. Inverse reconstruction

Forward modeling Forward modeling • Electromagnetic head model Reconstruct electrode signals from electrical • Reconstruct electrode signals from electrical current in the head

Head model Head model Spherical approximation Spherical approximation

Realistic head model Realistic head model

• Boundary element method • Finite element method

SPM head model SPM head model Compute transformation T

Individual MRI Templates

Apply inverse transformation T‐1

Individual mesh Individual mesh

BEM mesh

Head model Head model • Electrode locations Electrode locations • Registration  – Landmark based L d kb d – Surface matching 

fiducials

• Leadfield fiducials

Rigid transformation (R,t)

Individual sensor space

Individual MRI space

Inverse approaches Inverse approaches Dipole

Distributed dipoles Distributed dipoles

Least‐square or Beamforming

More unknowns than data

Distributed approach Distributed approach • Y = KJ+ E • No unique solution! – Priors:            min( Pi i ( ||Y – KJ||2 + λf(J) ) • minimum overall activity • Location • Smoothness

• Bayesian model comparison Bayesian model comparison

References Sylvain Baillet Sylvain Baillet’ss presentation at HBM 2008 HBM 2008 SPM for dummies 0000‐2008 presentations http://www bci2000 org http://www.bci2000.org Baillet et al., IEEE Sig. Proc. Mag., 2001 M tt t Philli & F i t (2005) SPM course Mattout, Phillips & Friston (2005) SPM http://www.fil.ion.ucl.ac.uk/spm/course/slide s05/ppt/MEEG_inv.ppt  05/ t/MEEG i t • SPM manual • • • • •