Anatomically and Functionally Constrained MEG/EEG Source Estimates - - PowerPoint PPT Presentation

Matti Hämäläinen

MGH/MIT/HMS Athinoula A. Martinos Center for Biomedical Imaging Charlestown, MA USA

Anatomically and Functionally Constrained MEG/EEG Source Estimates

Matti Hämäläinen 7/2009

Contents

• Introduction: Multimodal imaging, MEG and EEG
• Anatomically and functionally constrained source

estimates

• Recent developments

2

Matti Hämäläinen 7/2009

Noninvasive Multimodal Brain Imaging

3 MEG MRI Anatomy

Estimated neural currents

EEG fMRI NIRS

MEG and EEG

Matti Hämäläinen 7/2009

MEG and EEG

V EEG MEG Jp Jp V EEG Jp

• The primary current is related

to the postsynaptic activity

• The primary current generates a

potential distribution (EEG) and the associated volume currents

• The primary and volume currents

together also create a magnetic field (MEG)

• However, the net effect of volume

currents is rather straightforward to take into account in MEG whereas the it is difficult to compute the EEG potential distribution accurately

Matti Hämäläinen 7/2009

Realistically-shaped forward models for MEG and EEG

6

Homogeneous model: skull taken as an insulator, result independent of conductivity Multilayer model: skull and scalp taken into account, conductivities needed

MEG ≈ EEG ≠

Theoretical analysis: Hämäläinen and Sarvas, 1989 Experimental validation: Okada et al., 1999

Matti Hämäläinen 7/2009

Primary currents in the cortex

7

Primary currents

MEG = 0 MEG 0 EEG 0 EEG 0 cortex current sources

B = 0

No magnetic field from radial currents in the sphere model

Matti Hämäläinen 7/2009

Tangential, radial, and tilted sources

8

EEG

tangential radial tilted

MEG

tangential radial tilted MEG has only one prototypical field pattern

Matti Hämäläinen 7/2009

MEG and EEG sensitivity to cortical sources

9

MEG EEG

Anatomically and functionally constrained source estimates

Matti Hämäläinen 7/2009

Motivation to use distributed source models

• Account for non-focal sources
• Automatic analysis without heuristic choices often

needed in multidipole models

• Incorporate anatomical and functional MRI

constraints

• Lower SNR data can be processed

Transformation of data to brain space without strong assumptions about the sources

11

Matti Hämäläinen 7/2009 12

Minimum-Norm Solutions Data and noise: Fit the data with a source penalty term: – Minimum-norm estimates (MNE): p = 2 – Minimum-current estimates (MCE): p = 1

ˆ q = argminq

• ||y − Gq||2

C + ||q||p R

• y = Gq +

E{T } = C

Matti Hämäläinen 7/2009

Retinotopic mapping with MNE

13 Ahlfors et al. 1992

฀฀฀฀฀฀฀

฀

Matti Hämäläinen 7/2009

Modern MNE

14

• Source locations (and orientations) constrained to

the cortical mantle

• Forward solution with BEM
• Full noise-covariance matrix estimates computed

from raw data

• Display on an inflated cortex to reveal the sulci
• Compute statistics
• Combined MEG and EEG solutions
• fMRI-guided solutions

Dale et al. 2000

Matti Hämäläinen 7/2009

Cortical Source Location Constraints

15

Tessellation of the cortex: Source location and orientation information For source estimation, the surface is typically decimated, resulting in 6000 - 10000 source locations

Matti Hämäläinen 7/2009

Inflated Cortex

16

Dale, Fischl, Sereno et al.

Topologically correct tessellation can be inflated

Matti Hämäläinen 7/2009

Inflation to a Sphere and Registration

17

Individual Aligned with average brain Align sulcal patterns to the average brain Mapped to the average brain Morph MEG activity estimate

Matti Hämäläinen 7/2009

Noise normalization

• Convert the current values into a test statistic

– dSPM (Dale et al.) – sLORETA (Pascual-Marqui et al.)

• Divide the current with its standard deviation
• Analyze MEG/EEG data like fMRI or PET

18

Dale et al. 2000

Matti Hämäläinen 7/2009

MNE and dSPM

19

MNE dSPM

• Auditory MEG data
• Source locations constrained to the cortex
• No orientation constraint
• dSPM and sLORETA produce very similar results with real data

Matti Hämäläinen 7/2009

Loose orientation constraint

20

• Penalize current components tangential to the

cortex

• Takes the finite spacing between elementary

sources into account

Lin et al. 2006

MEG = 0 MEG 0 EEG 0 EEG 0 cortex current sources

Matti Hämäläinen 5/2009

Effect of additional constraints

21

Unweighted Depth-weighted Depth-weighted and LOC MNE dSPM

Location of simulated source

Matti Hämäläinen 7/2009

Spatial dispersion of cortically-constrained MEG and EEG source estimates

MEG EEG MEG+EEG

0 cm 4 cm 2 cm Molins et al. 2008

Matti Hämäläinen 7/2009

Comparison of MEG, EEG, and fMRI (dSPM)

23

fMRI MEG EEG MEG + EEG

Sharon et al. 2007

Matti Hämäläinen 7/2009

Visual percepts of an ambiguous scene

24 Parkkonen et al., PNAS, 2008 10 10 20 20 30 30

MEG signals at an occipital sensor

Spectral density (fT/cm/ Hz)

• 12 Hz

15 Hz

10 Hz 12 Hz 15 Hz 20 Hz

Time / s

Percept ‘faces’ Percept ‘vase’

• 1
• 2

1 2

12 Hz 15 Hz Noise:

Matti Hämäläinen 7/2009

Extract tag-related activity: MNE + GLM

25 Parkkonen et al., PNAS, 2008

Behavioral report Source waveform GLM

= = * *

0.25 s T T T T 1 s ’faces’ ’vase’ tags 12.0 Hz 15.0 Hz alpha 9.5 Hz mu 10.0 Hz 18.9 Hz mains 50.0 Hz linear trend aV aF bV bF t > 2 s

Matti Hämäläinen 7/2009

Group analysis

26 'vase' 'faces' Amplitude ratio vase-tag / face-tag Percept 2.0 1.5 1.0

N = 8 Left Right

100% 75% 50%

Parkkonen et al., PNAS, 2008

Significant activity in either tag frequency Amplitude ratio at ROI

Matti Hämäläinen 7/2009

fMRI-guided estimates

Dale et al. 2000

• Prioritize locations of significant fMRI activity

(increase source variance)

• fMRI incorporated as a constraint, not an

integrated analysis procedure

27

Matti Hämäläinen 7/2009

Time:

• Region-specific, independent neural and vascular time cources

Space:

• Time courses modulated by a scalar z
• z is location specific and smooth in space

28

Going further: the fIRE Model

Ou et al. 2009

fMRI-Informed Regional Estimates

Matti Hämäläinen 7/2009

FIRE: Graphical Model

29 Brain activity, space Current sources, space & time fMRI data, space & time EEG/MEG data, space & time Neural waveform, time Vascular waveform, time

Ou et al. 2009

Matti Hämäläinen 7/2009

Scenario One: No Silent Sources

25

Simulation MNE fMNE fARD FIRE

MNE fMNE fARD FIRE Simu.

Matti Hämäläinen 7/2009

Scenario Two: Silent Vascular Activity

26

MNE fMNE fARD FIRE Simu.

Simulation MNE fMNE fARD FIRE

Matti Hämäläinen 7/2009

Scenario Three: Silent Neural Activity

27

MNE fMNE fARD FIRE Simu.

Simulation MNE fMNE fARD FIRE

Matti Hämäläinen 7/2009

FIRE: Features and Benefits

33

• Avoids excessive bias towards fMRI

– If the hemodynamic signal is missing, defaults to the L2 minimum-norm solution – If the neural signal is missing does not attempt to imply neural signals at the fMRI-only regions

• Regional approach: computationally tractable

Matti Hämäläinen 7/2009

Multimodal imaging involving MEG and EEG

• MEG and EEG have complementary strengths
• Anatomical MRI

– Visualization in anatomical context – Modeling constraints for improved accuracy and to make the source estimation problem less ill posed – Cortical surface visualization common with fMRI: easy comparison, common group analysis approaches

• Functional MRI

– Comparison of data in the same format – fMRI-informed MEG/EEG estimates to enhance spatial resolution – Goal: Integrated analysis

34

Matti Hämäläinen 7/2009 35

Thank you!

Seppo Ahlfors Jack Belliveau Anders Dale (UCSD) Bruce Fischl Polina Golland (MIT/CSAIL) Riitta Hari (HUT) Fa-Hsuan Lin Maria Mody Antonio Molins Wanmei Ou (MIT/CSAIL) Lauri Parkkonen (HUT) Tommi Raij Bruce Rosen Dahlia Sharon (Stanford) Daniel Wehner Thomas Witzel www.nmr.mgh.harvard.edu