Session 2 - BOLD imaging Flashcards
(24 cards)
MRI: what is needed?
- magnet
- gradient coil
- radio frequency coil
Early research
Angelo Mosso (1846-1910):
- tilt table: if a participant needs to complete a certain mental task, the table tilts a bit due to increased blow flow to the brain
fMRI: what is it/what can be learnt?
= functional magnetic resonance imaging
- in MRI: proton and proton spin
- in fMRI: still proton and proton spin but additionally: magnetic properties of haemoglobin and neurovascular coupling
–> T2* is influenced by BOLD effect
MRI sequences
T1: slow recovery of longitudinal magnetisation due to spin-lattice interactions
T2: fast dephasing of spins due to spin-spin interactions
T2*: fast dephasing of spins due to combination of spin-spin interaction and magentic field inhomogeneities –> influenced by BOLD effect
BOLD Signal
= Blood Oxygen Level Dependent Signal
- changes between oxygenated and de-oxygenated blood
–> proton spin depends on oxygenation - oxygenated blood (-) is less magnetic (diamagnetic) than de-oxygenated blood (+) (paramagnetic)
Rapid T2* dephasing of deoxygenated Hb
de-oxy-Hb dephases more rapidly:
- less O2
- stronger magnet (paramagnetic)
- stops faster
- less MRI signal (=darker)
oxy-Hb dephases more slowly:
- more O2
- weaker magnet (diamagnetic)
- runs longer
- more MRI signal (=brighter)
less de-oxy-Hb/more oxy-blood –> more T2* signal, longer T2*, brighter
more de-oxy-blood/less ocxy-blood –> shorter T2, less T2 signal, darker
Magnetic properties of blood
MRI DOES NOT measure haemoglobin or oxygen but the spin properties of protons: changes when haemoglobin looses oxygen
- blood appears dark in comparison to water: de-oxy-Hb is overall darker than oxy-Hb (brighter centre)
- signal is caused by susceptibilty changes between oxygenated and de-oxygenated blood –> proton spin changes depending on level of blood oxygenation level
How does the BOLD signal change during increased brain activity?
Original idea:
- more neural activity –> less oxygen, darker image
- increase in neural metabolism leads to increases in de-oxy-Hb –> hydrogen spin relaxation faster, T2* decreases
BUT NOT TRUE: Neurovascular coupling
- increase of de-oxy-Hb but at the same time more oxy-Hb from the outside (much more than de-oxy!)
- normally there is a certain blood flow rate but in case of neural activity: overshoot
Neural metabolism increases
- increase in cerebral blood flow
- decrease in de-oxy-Hb and increase in oxy-Hb (even though oxygen is used up by neural activity)
- hydrogen spin relaxation slower, T2* increases
Early fMRI studies
- before fMRI always contrast agents needed
- first BOLD without contrast agent: 1992
- different experiments: tapping/visual checker board
- block or event-related design
- observed increases and decreases during and after the stimuli
Haemodynamic response funtion (HRF)
= temporally extended response to nerve firing
- initial dip (not visiable with all fMRIs, reflects possibly the dip in oxy-Hb)
- more focal - overshoot, increase of BOLD signal intensity (increased blood flow, more oxygenated blood)
- signal begins to rise soon after stimulus onset - peak
- signal peaks around 4-6 sec after stimulus onset - decay, decrease to baseline
- undershoot
- about 10-20s after stimulus onset
- post stimulus
- signal suppressed after stimulus ends
- takes about 20 to 30s to return to base line
baseline –> initial dip –> overshoot –> undershoot (–> baseline)
- signal change in % (point-baseline)/baseline –> usually 0.5-3%
- there is a link between neural activity and blood flow –> exact details unknown
spatial and temporal resolution
- MRI can reach good spatial resolution: ~1log mm
- but temporal resolution is lower: seconds to days
- structural MRI has higher spatial resolution (1x1x1mm) versus fMRI typical 3x3x3mm
What does a singel voxel (1mm^3) contain?
- 10^4-10^5 neuons
- 10^8-10^9 synapses
- 300m of dendrites
- 4000m of axons
- 0.4m of capillaries (in macaque visual crotex)
Partial volume effect
- in larger voxel: 3x3x3mm (27mm^3)
- less spatial resolution: more different tissues
Spatial precision: Monkey somatosensory cortex
- fMRI: 3x3x3mm (can be higher with higher field strength or better recording methods sMRI)
- not perfect but some overlap –> idea about where activity happens
- not exact location but precise enough
Temporal resolution
- typically ~2s apart (can be faster) but extended HRF response (20-30s)
- increasing sampling rate allows for more fine-grained sampling of HRF but does not directly allow to separate different events!
- TR (time to repeat): every 2 seconds but more often works too
- does not allow to separate events
Temporal precision
duration:
- to map trials against each other (+RT) –> get temporal effects
Onset latencies:
- precise even if we don’t know when the onset is
- in same reion & subject: relatively precise onset
- different for brain regions and subjects (maybe conditions too)
HRF variability
- differs between and within subjects
- no correlation between peak latency and amplitude
Summation of BOLD responses
- HRF increases from trial one to trial three
–> but if substracted from each other: always approx. the same shape
Linearity of BOLD response
- linearity is a good approximation (successfull in application)
- scaling and superposition
- convolution: canconical HRF*boxcar regression (ex. time of finger tapping) –> get estimation of what BOLD singal would look like (can be compared to actual measure: worked)
Non-linearity of BOLD
at <6s
- close but not perfect
Spatial precision
- submilimetre precision but might not completety agree with electrophysiology
Source of the signal?
hypotheses:
- firing of excitatory cells
- local inhibition
- firing of inhibitory cells
- feed-forward input
- lateral input (on ex. or inh. neurons)
- some complicated combination??
–> not important for measuring but maybe for precise interpretation of the signal
What does model the neurovascular coupling better LFP or MUA?
LFP = Local field potentials
- 40-130Hz
- reflects summation of post-synaptic potentials
MUA = Multi-Unit Activity
- 300-1500Hz
- reflects action potentials/spiking
–> slightly better correlated with/predicted by LFP (input) of a region
–> typically assumed to represent postsynaptic inhibitory and excitatory potentials
BOLD vs Electrophysiology
- Moving dot stimuli
- compare average monkey physiology to average BOLD signal in humans
- idea: same stimuli (length) same/different individuals: same activity expected
- movies in patients (with brain surgery, prep with long-term measurements) –> measures the same thing
