EEG data correction

Report
Simultaneous recording of EEG and BOLD responses
Why and How
Synopsis
1.
Motivation and perspectives
2.
Technical Setup
3. EEG data processing
i.
ii.
The gradient artifact
•
Technical prerequisites: synchronization
•
Artifact removal and data quality
The ballistocardiographic artifact
4.
Current studies
5.
Conclusions
Motivation and perspectives
• Achieving both high spatial and temporal resolution
• Shed light on the foundations and interrelations of MEG, EEG and fMRI
Motivation and perspectives
Is there a (partial) correspondence of fMRI and EEG/MEG?
• fMRI indirectly inferes neural activity via BOLD-reponse (neurovascular
coupling)
• EEG/MEG more directly reflect neural activity (apical EPSPs…)
large scale
synchrony
neural firing
rates
Motivation and perspectives
Basic applications
• fMRI-informed source reconstruction
• parametric designs and EEG-fMRI covariation
• single-trial coupling of EEG and fMRI
Motivation and perspectives
Higher order models
• compound neural mass and hemodynamic models
• joint ICA
• parallel ICA
Motivation and perspectives
Clinical relevance???
Original Motivation:
• Mapping epileptic zones
Recent „clinical“ research:
• Movement disorders (cortical myoclonus)
• Brain-computer interfaces (Biofeedback)
Motivation and perspectives
Measurement techniques and applications
• separate recordings of EEG and fMRI (two sessions)
• interleaved recordings (EEG in “silent periods”)
• simultaneous recordings (both modalities continuously
measured)
Motivation and perspectives
Continuous/simultaneous measurements:
Fp1
Fp1
12
1010
µV
µV
Fp2
Fp2
• temporal correlation of EEG and fMRI
F3
F3
F4
F4
C3
C3
C4
C4
10
P3
P3
P4
P4
• avoidance of order effects
O1
O1
O2
O2
F7
F7
8
F8
F8
• semi-optimized design
T7
T7
T8
T8
P7
P7
P8
P8
Cz
Fz
Fz
6
Cz
Cz
• strongly degraded signal quality (especially EEG)
Pz
Pz
Iz
Oz
FC 1
FFT9
4
FC 2 0
FFT1
C P1
TFT9
TFT1
C P2 0
TP9
FC 5
TP10
FC 6
2
C P5
PO9
PO10
C P6
FT1
TP9 1
FT1
2
TP10
EOG
F1
EC
F2 G
0
20
40
60
R128
R128
Sync
On Beta
Sync On SR128
4
Sync On
1 Sync On
64Sync On
2ThetaSync OnAlpha
1
1
1
Sub-Delta
Delta
80
100
Sync On 2 R128
S
Sync
1 On
64
contaminated
raw „clean“ EEG
EEG
120
Sync 1
On
R128
Sync64
On
Technical Setup
Combined EEG – fMRI
Recordings
Actual Status Hard- and Software
Technical Setup
EEG-Recording
System Components (BrainAmp MR plus, Brain Products GmbH):
1.
EEG amplifier unit, 32 channel, fMRI
approved (GE, Bruker, Siemens and
Phillips scanner), accumulator driven
2.
EEG cap (EASY Cap), 32 channel (plus EOG, ECG), modified 10-20 system,
sintered Ag/AgCl sensors, 10 kOhm for EEG cables, 15 kOhm for EOG/ECG
cables, 3 different sizes
3.
Sync-Box (Frequency divider), synchronization between MR scanner and
EEG data recording
4.
EEG-Data acquisition computer + Recording Software
5.
BrainAmp I/O USB Adapter,
interface between all other components
Technical Setup
EEG cap
Technical Setup
EEG Amplifier
Technical Setup
Stimulation Modes
1. Visual Stimulation:
Stimulation Computer (Presentation) -> Beamer -> Ground Glass ->
Mirror (800x600 pixel) -> Subject
2. Auditory Stimulation:
Stimulation Computer (Presentation) -> Audiometer -> Audio Amplifier
-> MR compatible stereo Head Phones -> Subject
3. Tactile Stimulation:
Stimulation Computer (Presentation) -> pneumato-tactile Stimulator ->
8 (finger) membranes -> Subject
Components which are inside the MR measurement chamber are emphasized in green
Technical Setup
Tactile Stimulation
• driven by compressed air
• up to eight independent output channels
• integrated TTL trigger control unit
Technical Setup
MRI compatible opto-electrical Response Unit
– 2 response panels (shape is adapted for left and right hand)
– Each panel provides 2 response buttons (best fitting for index and
middle finger)
– Response panels are connected to opto-electrical transducers
via fiber optical cables (inside MR chamber)
– Response signals are recorded by Stimulation and Recording
Software in order being referable during later analysis
Technical Setup
Response Unit
Technical Setup
Triggering / Synchronization
(Hardware) Trigger Generators:
1. Stimulation Computer: event coding and timing via Presentation
port codes
2. Response Unit: response coding trigger
3. SyncBox: periodic sync trigger generated from scanner electronic
pulse to synchronize the EEG signal sampling by the MR scanner
rate (requisite for scanner artefact rejection)
4. fMRI-Scanner: volume trigger representing MR volume scan onset
time (used for scanner artefact rejection and event timing in
Presentation)
All triggers are represented in the recorded EEG data set and one can refer to
them during the subsequent data analysis (artefact rejection, averaging etc.).
Technical Setup
Beamer
fMRI
Scanner
Electronic
EEGAmplifier
MR chamber
Head Phones
Volume Trigger
Sync
preAmp
Audio
Amplifier
Opto-elect
Transducer
Pneumatotactile
Stimulator
Response
Buttons
Clips
Membranes
I/O-USB
Adapter
Sync Box
EEG
Recording
Stimulation
Technical Setup
Online Recording Setup
Technical Setup
Combined EEG – fMRI
Recordings
Data quality
EEG data correction
Major artifacts
• “gradient artifact”
• induced currents due to gradient switching
• “ballistocardiographic artifact”
• movement of conductive material in static magnetic
field
• vibrations due to active helium pump
EEG data correction
The “gradient artifact”
• slice selection:
• frequency of slice acquisition
• e.g. TR = 2s, 28 slices – 14 Hz (and harmonics)
• spatial encoding within a slice:
• usually phase encoding
• e.g. 64 × 64 Matrix – 64 × 15 = 960 Hz (not recorded)
EEG data correction
The “gradient artifact”
• technical artifact – rather invariant
• correction via subtraction of channel-specific templates
• problem 1: subject motion changes position of
cables/electrodes
• foam cushions
• problem 2: differential timing of EEG sampling and fMRI
acquisition
• EEG/MR Synchronisation – “SyncBox”
EEG data correction
synchronized
unsynchronized
EEG data correction
Fp1 12
12
12
10 µV
Fp2
F3
F4
C3
C4
10
10
10
P3
P4
O1
O2
F7
88 8
F8
T7
T8
P7
P8G
EC
Cz
Cz
Fz
66 6
Cz
Pz
Iz
FFT9
FFT1044 4
TFT9
TFT10
TP9
TP10
PO9
22 2
PO10
FT11
FT12
EOG
ECG
00 0
20
2020
R128
Sync Delta
On
Sub-Delta
Sub-Delta
Sub-Delta
Delta
Delta
Sync Theta
On
Theta
Theta
40
4040
Sync On
Sync On
Alpha
Alpha
Alpha R128
Beta
Beta
Beta
60
6060
Sync On
80
8080
Sync On
R128
Sync On
100
100
100
Sync On
120
120
120
Sync On
contaminated
raw
corrected
EEG
EEG
EEG
corrected EEG
with„clean“
sluggishly
fixed electrode
R128
Sync On
EEG data correction
EEG data correction
The ballistocardiographic artifact
• “ballistocardiographic artifact”
• movement of conductive material in static magnetic
field
a) cardiac-related axial head motion
b) pulsatile movement of the scalp
c) electromagnetic induction due to blood flow
EEG data correction
The ballistocardiographic artifact
• correction via subtraction of channel-specific templates
Problems:
• biological artifact – high degree of variability
• template stability over time – motion induced changes
EEG data correction
R
R
Fp1
R
200
µV
200
10 µV
Fp2
F3
F4
R
R
R
C3
R
R
C4
R
P3
P4
R
O1
O2
F7
F8
T7
T8
P7
P8
Fz
Cz
Pz
Iz
FFT9
FFT10
TFT9
TFT10
TP9
TP10
PO9
PO10
FT11
FT12
EOG
R
R
R
R
R
R
R
R
ECG
Sync
S On
2 S
R128
32
Sync On
Sync On
R128
Sync On
R
R
R
R
R
R
R
R
Sync
S 4
On
R128
Sync On
R
R
R
Sync On
S 1 R128
Sync On
BCG artifact –BCG
afterartifact
template subtraction
Sync On
R
R128
Sync On
EEG data correction
EEG data correction
The ballistocardiographic artifact
• further improvements may be obtained via:
• removal of residual BCGA via ICA
• Optimal Basis Set (OBS – channelwise temp. PCA)
• OBS - ICA
EEG data correction
Fp1
200 µV
Fp2
F3
F4
C3
C4
P3
P4
O1
O2
F7
F8
T7
T8
P7
P8
Fz
Cz
Pz
Iz
FFT9
FFT10
TFT9
TFT10
TP9
TP10
PO9
PO10
FT11
FT12
EOG
R
R
R
R
ECG
Sync
S On
2 S
R128
32
Sync On
Sync On
R128
Sync On
R
R
R
R
R
Sync
S 4
On
R128
Sync On
Sync On
S 1 R128
Sync On
BCG
ICA filtering
BCGartifact
artifact––after
afteradditional
template subtraction
Sync On
R
R128
Sync On
EEG data correction
12
12
10
10
8
8
Cz
Cz
6
6
4
4
2
2
0
0
Sub-Delta
Sub-Delta
20
20
Delta
Delta
40
40
Theta
Theta
Alpha
Alpha
60
60
80
80
100
100
Beta
Beta
BCG
ICA filtering
BCGartifact
artifact––after
afteradditional
template subtraction
120
120
EEG data correction
The ballistocardiographic artifact
• further improvements may be obtained via:
• removal of residual BCGA via ICA
• Optimal Basis Set (OBS – channelwise temp. PCA)
• OBS – ICA
• “automatized” component identification
• correlating the raw ECG-trace with time courses of
independent component
• correlating BCGA-topography with IC weighting matrix
EEG data correction
F00
10
5 µV
µV
F01
F02
F03
F04
Fp1
F05
F06
1000 µV
F07
F08
F09
F10
F11
F12
F13
F14
F15
EOG
F16
F17
1000 µV
F18
F19
F20
F21
F22
F23
F24
R
R
F25
R
R
R
R
R
R
R
R
R
R
R
R
F26
ECG
F27
F281000 µV
F29
F30
F31
R
R
R
R
R
R
R
R
R
R
S Sync
Sync
S2 2Sync
On
On
SR128
On
SR128
2S
32
Sync
S128
32
Sync
S128
S
R128
32
Sync
On
On OnSync
Sync
Sync
On
OnR128
On
R128
SSync
ync
R128
Sync
On
On
SS 8
On
8 Sync
SyncSync
On
On
S 4
R128
On
R128
S
Sync
ync
R128
Sync
On
On On
S
S Sync
Sync
22
Sync
On
On
S
S 32
R128
32
R128
On
S Sync
1
Sync
R128
Sync
On
On OnSync
SyncSync
On
On R128
R128
On
S Sync
8 R128
Sync
On On
EEG data analysis
Fz
Fz
10 µV
10 µV
50 µV
248 ms
-7.84 µV
0.00 µV
240 ms
344 ms
7.84 µV
-4.83 µV
0.00 µV
348 ms
4.83 µV
-15.00 µV
-14.32 µV
0.00 µV
0.00 µV
15.00 µV
14.32 µV
Cz
Cz
10 µV
10 µV
Pz
Pz
10 µV
10 µV
subtraction only
additional ICA filtering
EEG data analysis
EEG data analysis
amplitude
FCz
time
-15.00 µV
0.00 µV
-15.00 µV
15.00 µV
0.00 µV
15.00 µV
EEG data analysis
standard fMRI
single trial fMRI
Conclusions
Current studies:
• Tactile Stop-Signal task (executive functions)
• Affective conditioning
• Language processing
Planned study:
• Resting state/default mode network

similar documents