...

MEASURING BRAIN FUNCTION IN BRAIN DEATH, COMA

by user

on
Category:

therapy

8

views

Report

Comments

Transcript

MEASURING BRAIN FUNCTION IN BRAIN DEATH, COMA
Impaired top-down processes in the vegetative
state revealed by SPM analysis of EEG data
Mélanie Boly, MD, PhD
Wellcome Trust Centre for Neuroimaging,
Functional Imaging Laboratory,
University College London
Coma Science Group
Cyclotron Research Centre
& Neurology Department
CHU Sart Tilman, Liège, Belgium
www.comascience.org
introduction | scalp level analysis| DCM | conclusion
Altered
states of consciousness
Consciousness
Conscious
Wakefulness
Locked-in syndrome
Drowsiness
REM
Sleep
Deep Sleep
Light sleep
Minimally Conscious State
40 % misdiagnosis!
General
Anesthesia
Coma
Laureys & Boly, Current Opinion in Neurology 2007
Laureys & Boly, Nature Clinical Practice 2008
Vegetative state
Somnambulism
Epilepsy
Schnakers et al., BMC Neurology 2009
introduction | scalp level analysis| DCM | conclusion
Diagnosing
consciousness: the challenge
Consciousness
Conscious
Wakefulness
Locked-in syndrome
Drowsiness
REM
Sleep
Deep Sleep
Light sleep
Minimally Conscious State
General
Anesthesia
Coma
Functional
neuroimaging
Vegetative state
Somnambulism
Epilepsy
Boly, Massimini & Tononi, Progress in Brain Research 2009
Boly, Current Opinion in Neurology, in press
Neural correlates of
consciousness (NCC)
Auditory NCC
Boly et al., Archives of Neurology
2004
VS
MCS
subliminal
?
preconscious
conscious
Di et al., Neurology 2007
Dehaene et al., TICS 2006
Diatz et al., JCognNsci 2007
introduction | scalp level analysis| DCM | conclusion
NCC in healthy volunteers
Garrido et al., PNAS 2007
Garrido et al., Neuroimage 2008
Del Cul et al., PLOS Biol 2007
Best correlate of conscious perception = long latency ERP components
Suggested involvement of backward connections in their generation
introduction | scalp level analysis| DCM | conclusion
MMN design – roving paradigm
Garrido et al., Neuroimage 2008, 2009
Scalp level analysis
www.comascience.org
introduction | scalp level analysis| DCM | conclusion
ERP data analysis – Methods
22 controls, 13 MCS and 8 VS patients
EEG data:
60 electrodes EEG acquisition system (Nexstim) – 15 min acquisition
Sampling rate 1450 Hz
~200 standard, 200 deviants per subject
CT scan or structural MRI obtained for each subject
Boly, Garrido et al., Science 2011 in press
introduction | scalp level analysis| DCM | conclusion
ERP data analysis – Methods
22 controls, 13 MCS and 8 VS patients
EEG data:
60 electrodes EEG acquisition system (Nexstim) – 15 min acquisition
Sampling rate 1450 Hz
~200 standard, 200 deviants per subject
CT scan or structural MRI obtained for each subject
SPM data analysis:
High pass filtering 0.5 Hz
Low pass filtering 20 Hz (to decrease EMG-related noise in the signal)
Downsampling at 200 Hz
Correction for ocular artifacts (Berg method from SPM) on continuous signal
Epoching -100 to 400 ms
Averaging data at the single subject level – standard & deviant (11th repetition) conditions
Convert to images in SPM
Boly, Garrido et al., Science 2011 in press
introduction | scalp level analysis| DCM | conclusion
ERP data analysis – Methods
22 controls, 13 MCS and 8 VS patients
EEG data:
60 electrodes EEG acquisition system (Nexstim) – 15 min acquisition
Sampling rate 1450 Hz
~200 standard, 200 deviants per subject
CT scan or structural MRI obtained for each subject
SPM data analysis:
High pass filtering 0.5 Hz
Low pass filtering 20 Hz (to decrease EMG-related noise in the signal)
Downsampling at 200 Hz
Correction for ocular artifacts (Berg method from SPM) on continuous signal
Epoching -100 to 400 ms
Averaging data at the single subject level – standard & deviant (11th repetition) conditions
Convert to images in SPM
Random effects analysis – 3 groups x 2 conditions
Patient’s prognosis entered as a covariate of no interest
F test for differential response to standard versus deviants in each group
F test for an effect of consciousness level on the amplitude of this response
Threshold FWE corrected p<0.05 at the voxel level
Boly, Garrido et al., Science 2011 in press
introduction | scalp level analysis| DCM | conclusion
MMN results – scalp level
RESPONSE TO DEVIANTS
Controls
introduction | scalp level analysis| DCM | conclusion
MMN results – scalp level
RESPONSE TO DEVIANTS
Controls
MCS
introduction | scalp level analysis| DCM | conclusion
MMN results – scalp level
RESPONSE TO DEVIANTS
Controls
MCS
VS
introduction | scalp level analysis| DCM | conclusion
MMN results – scalp level
RESPONSE TO DEVIANTS
Controls
MCS
VS
introduction | scalp level analysis| DCM | conclusion
MMN results – scalp level
RESPONSE TO DEVIANTS
Controls
MCS
VS
introduction | scalp level analysis| DCM | conclusion
MMN results – scalp level
Boly, Garrido et al., Science 2011 in press
introduction | scalp level analysis| DCM | conclusion
MMN results – scalp level
Boly, Garrido et al., Science 2011 in press
introduction | scalp level analysis| DCM | conclusion
MMN results – scalp level
RESPONSE TO DEVIANTS
Correlation between the level of consciousness and:
- Global amplitude of the ERP response
- Predominant late components in latency of ERP
- Involvement of frontal topography at the scalp level
Connectivity analysis
using DCM
www.comascience.org
introduction | scalp level analysis| DCM | conclusion
DCM for EEG - principles
Explain a given M/EEG signal at the neuronal level
Which brain network
creates this ERP?
And how?
introduction | scalp level analysis| DCM | conclusion
MMN design – roving paradigm
Garrido et al., Neuroimage 2008, 2009
introduction | scalp level analysis| DCM | conclusion
DCM for EEG - principles
Electromagnetic forward model for M/EEG
Depolarisation of
pyramidal cells
x0  f ( x, u, )
Forward model:
lead field & gain matrix
LK
Scalp data
y  g ( x, )  LKx0
Forward model
www.comascience.org
Spatial Forward
Model
Depolarisation of
pyramidal cells
x  f ( x, u,  )
Spatial model
L
L

Sensor
data
y  L x0  g ( x0 ,  )
L
Default: Each area that
is part of the model is
modeled by one
equivalent current
dipole (ECD).
Neural mass model of a cortical macrocolumn =
CONNECTIVITY ORGANISATION
E
x
t
r
i
n
s
i
c
i
n
p
u
t
s
Excitatory
Interneurons
Pyramidal
Cells
POPULATION DYNAMICS
Function P
MEG/EEG
signal
mean firing
rate

mean
postsynapt
ic potential
(PSP)
Function S
mean PSP

mean firing rate
Inhibitory
Interneurons
Excitatory connection
Inhibitory connection
Between-area
connectivity
Inhibitory
IN
1
2
2
3
Excitator
y IN
1
Pyramidal
cells
Intrinsic
Forward
Extrinsic Backward
Lateral
Input u
David and Friston, 2003
David et al., 2005
introduction | scalp level analysis| DCM | conclusion
for EEG –fit
principles
Model DCM
Inversion:
the data
Data
Observed (adjusted) 1
6
4
2
0
We need to estimate the
extrinsic connectivity
parameters and their
modulation from data.
-2
-4
input
-6
-8
0
50
100
150
time (ms)
200
250
Predicted data
Predicted
6
4
2
0
-2
-4
-6
-8
0
50
100
150
time (ms)
200
250
introduction | scalp level analysis| DCM | conclusion
DCM for EEG - principles
Balance between
model fit &
model complexity
Alternative Models for Comparison
introduction | scalp level analysis| DCM | conclusion
DCM for EEG – group analysis
LD
MOG
FG
LD|LVF
MOG
FG
LD|RVF
MOG
LD|LVF
LG
-25
-20
LG
RVFLD|RVFLVF
stim.
stim.
LVF
stim.
Subjects
-30
MOG
LD
LG
m2
-35
FG
LD
LG
RVF LD
stim.
FG
-15
-10
m1
-5
0
Group level random effects BMS
resistant to outliers
5
Log model evidence differences
Stephan et al. 2009
introduction | scalp level analysis| DCM | conclusion
Bayesian model comparison
Boly, Garrido et al., 2011
introduction | scalp level analysis| DCM | conclusion
Bayesian model comparison
Boly, Garrido et al., 2011
introduction | scalp level analysis| DCM | conclusion
Bayesian model comparison
Boly, Garrido et al., 2011
introduction | scalp level analysis| DCM | conclusion
Bayesian model comparison
Boly, Garrido et al., 2011
introduction | scalp level analysis| DCM | conclusion
Bayesian model comparison
Boly, Garrido et al., 2011
introduction | scalp level analysis| DCM | conclusion
DCM – quantitative connectivity analysis
Boly, Garrido et al., 2011
introduction | scalp level analysis| DCM | conclusion
DCM – quantitative connectivity analysis
Impairment of BACKWARD connection from frontal to temporal cortices
is the only significant difference between VS and controls
*
ns
Ctrls
(p = 0.012)
*
MCS
(p = 0.006)
VS
Boly, Garrido et al., 2011
introduction | scalp level analysis| DCM | conclusion
DCM – quantitative connectivity analysis
Impairment of BACKWARD connection from frontal to temporal cortices
is the only significant difference between VS and controls
CONTROLS/MCS
VS
3
3
1
1
2
2
introduction | scalp level analysis| DCM | conclusion
DCM – quantitative connectivity analysis
Impairment of BACKWARD connection from frontal to temporal cortices
is the only significant difference between VS and controls
VS
3
1
2
Del Cul et al., PLOS Biol 2007
Conclusions
www.comascience.org
introduction | scalp level analysis| DCM | conclusion
Conclusion
SCALP LEVEL:
Correlation between response amplitude (latency >100 ms, involving frontal component) with
the level of consciousness
Boly, Garrido et al., Science 2011 in press
introduction | scalp level analysis| DCM | conclusion
Conclusion
SCALP LEVEL:
Correlation between response amplitude (latency >100 ms, involving frontal component) with
the level of consciousness
DCM ANALYSIS:
- Selective impairment in backward connectivity from frontal to temporal cortices in VS
- MCS patients show a pattern similar to controls
Fits very well with NCC in healthy volunteers (though only indirect evidence there for backward
processes being important beforehand)
First direct demonstration of a link between preserved top-down processes and the level of
consciousness in these patients
Future studies on a larger patient population to assess diagnostic utility and prognostic value
Boly, Garrido et al., Science 2011 in press
introduction | scalp level analysis| DCM | conclusion
Conclusion
SCALP LEVEL:
Correlation between response amplitude (latency >100 ms, involving frontal component) with
the level of consciousness
DCM ANALYSIS:
- Selective impairment in backward connectivity from frontal to temporal cortices in VS
- MCS patients show a pattern similar to controls
Fits very well with NCC in healthy volunteers (though only indirect evidence there for backward
processes being important beforehand)
First direct demonstration of a link between preserved top-down processes and the level of
consciousness in these patients
Future studies on a larger patient population to assess diagnostic utility and prognostic value
Hierarchy of
brain connectivity
Impairment in
unconsciousness
?
functional
Boly, Current Opinion in Neurology, in press
Buckner et al., J Neurosci 2009, Hagmann et al., PLOS Biology 2008
structural
University of Liège
Steven Laureys
Olivia Gosseries
Caroline Schnakers
Marie-Aurélie Bruno
Pierre Boveroux
Audrey Vanhaudenhuyse
Didier Ledoux
Jean-Flory Tshibanda
Quentin Noirhomme
Remy Lehembre
Andrea Soddu
Athena Demertzi
Rémy Lehembre
Christophe Phillips
Pierre Maquet
Stanford University
Michael Greicius
University of Cambridge, UK
Adrian Owen
Martin Coleman
John Pickard
Martin Monti
University of Milan
Marcello Massimini
Mario Rosanova
Adenauer Casali
Silvia Casarotto
University of Wisconsin - Madison
Giulio Tononi
Brady Riedner
Eric Landsness
Michael Murphy
Fabio Ferrarelli
Marie-Curie University, Paris
Louis Puybasset
Habib Benali
Giullaume Marrelec
Vincent Perlbarg
Melanie Pellegrini
Cornell University, NY
Nicholas Schiff
JFK Rehabilitation Center, NJ
Joseph Giacino
University College London, UK
Karl Friston
Marta Garrido
Vladimir Litvak
Rosalyn Moran
We thank the participating patients and their families
www.comascience.org
Any questions?..

www.comascience.org
Fly UP