Multimodal Brain Imaging Flashcards
what is Single Neuron Activity and Local Field Potential?
single -cell and ( Local Field Potential 140-100,00 neurons - so slightly better spatial resolution EEG)
so AP and PSP
for a Continuous Signal the sampling freq is?
(sampling frequency > 8 kHz)
settings for single cell?
- High pass filter (> 300 Hz)
* Spike sorting
details for LFP
Local Field Potential • Down-sampling (not necessary for lower sampling frequency) • Spike correction (local interpolation around each spike) • Low pass filter (< 200 Hz)
what was first finding in slides using these methods?
Neuronal Spikes Phase Locked to LFP
one neuron fired more at theta peaks and not troughs
one neuron fired more at theta troughs and not peaks
= phase preference
also - different neurons show diff preferences but all like theta
- Neurons also exhibited phase-locked to oscillations in frequencies other than theta, like delta (1-4 Hz), alpha (10-16 Hz), beta (16-30 Hz), and gamma (30-90 Hz) frequency ranges
- Some neurons show phase locking to multiple oscillations
what does this give us
more potential for coding information - like a large number of words in vocabulary
- Neuronal phase-locking phenomenon is present in various brain regions
- Gamma oscillations facilitate a different type of information coding compared with delta or theta oscillations
• Delta and Theta oscillations may facilitate phase-based temporal coding
– Phase Coding Hypothesis: Neurons encode information via the oscillatory phase at which they spike
> Different neurons fire at different phases of ongoing oscillations
when combing EEG and fMRI, the one issue is
different time scales (BOLD is sluggish)
3 different approaches to combined EEG and fMRI?
1) Separate EEG and fMRI - and subsequent combination
2) Simultaneous - recordings and analysis of high amplitude EEG signals (e.g. alpha oscillations)
3) Simultaneous EEG-fMRI recordings and analysis of ERP and single trials
Separate EEG-fMRI Advantage?
Better quality of signals because lesser artifacts
associated paper for separate?
Heinze et al (1994) - pet n EEG / auditory visual spatial attention + source localisation
Separate EEG-fMRI: Drawbacks ?
• Implicit assumption: Both measures pick up similar neural activity
– Not all activity in EEG result in BOLD responses and vice versa
- Data cannot be combined to investigate emergent brain responses which may not be time-locked to stimulus
- Significant subjective and experimental differences between the two recording sessions
• Other differences – Order effects
– Posture (EEG seated / fmri supine)
– Environmental noise – Level of comfort
Simultaneous EEG-fMRI Acquisition: the EEG is …
High amplitude EEG Signals
High amplitude EEG visible to the naked eye – Alphaoscillations
– Epilepticspikes
– SSVEP(steadystatevisualevokedpotential) At low MR field strengths (< 1.5 T)
associated Simultaneous paper =
Ives 1993
Ives 1993 found
Typical result:
– EEG Alpha power is negatively correlated with BOLD signalin occipital cortex and positively correlated with thalamic activity
Simultaneous needs
- Pre-amps inside MR magnet (reduction of cable length)
- Amps must be free of ferro-magentic materials
- Wider dynamic (range of input signal greater to get better recording - you dont want a saturated amplifier, gives no info) range to prevent amplitude saturation
What are the Two types of acquisition?
- Interleaved Acquisition
* Continuous Acquisition
in Interleaved Acquisition, EEG is acquired….
continuously
in Interleaved Acquisition, fMRI is acquired….
intermittently
in Interleaved Acquisition, following stimulus presentation…
1-2 seconds of EEG data are captured before a few seconds of fMRI data is aquired
This approach helps to reduce ….
interference from scanner noise
and is particularly useful in
auditory experiments
However, the main artifact xxx is still present
BCG - Ballistocardiogram / pulse artifact
but
it is less a problem as: BCG is most possibly not locked with stimulus-related response, so ERP is usually unaffected
what is BCG - Ballistocardiogram / pulse artifact?
Interference related to cardiac cycle and scales proportionately with the magnetic field strength
What are the Three possible sources of BCG?
• Small movements of the electrodes and the scalp due to expansion and contraction of scalp arteries between the systolic and diastolic phase
(micro chanages in electrode location)
- Fluctuation of the Hall voltage due to pulsatile changes of the blood in the arteries
- Small cardiac related movements of the body
Substantial variability within and between subjects, so its proper removal is the principal aim of getting a clean EEG signal
what is Fully Simultaneous Acquisition?
- Both EEG and fMRI are acquired continuously
* EEG amplifier must have a wider dynamical range so that the amplifiers do not saturate during fMRI acquisition
however… Fully Simultaneous Acquisition contains XXX and needs XXXX
- EEG signal contains large artifacts
- Sophisticated algorithms are necessary for artifact elimination (mainly on post-hoc basis)
- Very high sampling frequency (>10 kHz/channel) is required
(to record the artifact really well, in order to remove the artifact!!!! incredible)
(to prevent aliasing of scanner artifacts into the EEG signal and to detect and to remove gradient artifacts)
• More useful for clinical purpose (localizing seizure foci), but problematic for ERP study (low SNR)
3 main ways of removing BCG artifacts?
Removal of BCG artefact:
- Averaged artifact subtraction (AAS)
- Adaptive Filtering
- ICA techniques (independent components analysis)
what is AAS Procedure for BCG Artefact Removal?
- Detect the QRS peaks in the ECG (electrode on the chest) signal
- Align EEG activity time locked to these peaks
- The averaged template is then subtracted from the EEG
What is a gradient artifact?
Caused by rapidly switching magnetic field gradients during fMRI acquisitions
so to stop information loss through peaking - you use that siper braod amp to record it in order to reduce it
how do we Removal of Gradient Artifact (GA)?
Estimating magnetic gradients and subtracting the sum of these responses from the actual EEG
Estimating average artifact template
- Time averaging of noisy EEG sections to obtain a template
- Subtracting the template from the channel
- EEG amplifier must capture the large amplitudes introduced by the GA (high dynamic range)
- Accurate trigger timing is essential (high sampling frequency)
what are some EEG-fMRI: Task Design Issues?
• Event-related design may be better than blocked design
• Fast event-related designs are more used
– Shorter (1-7 sec) ITI (inter-trial-interval)
– Faster data acquisition
– Reduction of the impact of cognitive processes due to long ITI
– Performances of EEG near optimal (EEG is rapidly decorrelated)
– Successive BOLD responses can overlap but be reduced offline
what are the Performance characteristics of experimental fMRI design ?
– Estimation efficiency (the ability to make an estimate of BOLD) – Detection power (the ability to detect brain activation)
– Conditional entropy (measure of the randomness of the design)
the trade off is….
Block design offers good detection power but minimal estimation efficiency, whereas event-related design offers
maximum efficiency but poor detection power
two main methods to integrate data?
- Using ERPs as predictors of fMRI response
2. Dipole based biophysical modelling for integration
when Using ERPs as predictors of fMRI response, ….
• fMRI activation is correlated with the amplitude of a specific ERP component
• Group level analysis
– ERP amplitude and task-related contrast image per participant are subjected to a correlational analysis
Maps brain responses that are specifically correlated with a chosen ERP component
– No information about the temporal order of responses that led to the chosen ERP component
• Individual level analysis
– More powerful approach
– ERP amplitudes are used as covariates in a general linear model based analysis of the fMRI data
in fMRI Constrained Dipole Modeling…
the data can be single-trial or spontaneous EEG data, not necessarily time locked to an external event
• Clinical applications
– Epilepsy
– Sleep research
what is – Spike triggered fMRI?
fmri pulse is triggered by eeg potentials
what does resting - state fmri use?
continous recording - recall mind wandering / default mode network
early problem with TMS and EEG?
artifacts caused by strong magnet field from TMS - they got around it by shocking one hemi and recording the other
what tactic do we use these days?
sample and hold
what is sample hold?
Stimulus Artefact Suppressor
temp protect EEG system from the magnetic environment - TMS pulse is brief - so we briefly turn off EEG (few ms)
- Sample the signal
- Hold it constant during the stimulus
what do you also need to achieve this …
TMS Compatible EEG Amplifier - to electrically isolte the EEG system (temp shut down)
what artifacts in TMS / EEG?
- DC shift of the electrodes just under the coil, arising from the movement of the electrodes due to coil vibration
- Reflex activity from cranial muscles
• Unwanted auditory activation due to the coil click
(Bone conducted sound)
using ERP and TMS - Massimini showed TMS-evoked Response & State of Vigilance, what was the result?
Breakdown of Cortical Effective Connectivity in sleep
connectivity and cross talk when awake
TMS and Brain Oscillations
Thut! - always with the TMS
Posterior a-Oscillations and Visual Perception
tms to visual cortex = phosphene in vision = less alpha
Dynamic coupling between
tACS frequency and Effectiveness of the Stimulation = most likely to occur in the darkness alpha /// when light on, most likely in beta
BOLD does indeed correlate with …..
electrical neural activity
Summary
• BOLD certainly reflects an increase in neural activity (almost linear correlation)
- BOLD correlates more with LFP and especially with high frequency oscillations in the gamma range
- SNR of neural response is higher than SNR of BOLD (fMRI analysis probably underestimates the neural activity)