TASK 5 - fMRI Flashcards
structural imaging
= different types of tissue have different physical properties, used to construct static maps of physical structure of the brain
CT
= computerised tomography = amount of X-ray absorption in different types of tissue
- amount of absoption is related to tissue density
- exposure to small amount of radiation
- typically used in clinical settings
- cannot distinguish between grey matter and white matter + cannot be adapted for functional imaging
MRI
= magnetic resonance imaging = create images of soft tissue of the body; amount of water in each type of tissue varies
- different types of tissue behave differently when stimulated
- construct 3D image of the layout of tissues
- strong magnetic field is applied across part of body being scanned –> sends radiofrequency in (aligns protons) and collects output radiofrequency
Tesla
= strength of magnetic field of scanner
- the higher the Tesla (= the stronger the magnetic field of machine) –> the finer detailed the image
MRI
- physical basis
= uses the spinning protons of the hydrogen nuclei in water
- -> spinning: spin around north-south axis and they are oriented according to some direction
- hydrogen nuclei has only one proton with 1 water molecule on each side
- water is to some extent present in all tissue types
MRI
- precesssion
= what we use for measurement
- radiofrequency = speed of precession spins
- speed of precession is proportional to strength of magnetic field
MRI
- Lamar equation
= precession frequency (how fast the protons spin) is dependent on strength of magnetic field
- important to understand how images are formed
MRI
- alignment
= when radiofrequency pulse administered, the protons align and their magnetic fields add up
- all protons are synchronised
1. aligned to B0 (magnetic field of scanner) 2. aligned horizontally - pulse must fit to magnetic field, according to Lamar equation
- magnetic field stays in horizontal position for some time
MRI
- free induction decay
= FID = after each pulse, protons realign themselves with the magnetic field of the person
- produces an electromagnetic echo
- how fast a signal decays/dephases depends on homogeneity of magnetic field in the nighbourhood
relaxation times
- T1
= longitudinal relaxation time = time constant which determines the rate at which excited protons return to equilibrium
- measure of time taken for spinning protons to realign with external magnetic field (B0)
- -> both relaxation times (T1 + T2) depend on tissue type
relaxation times
- T2
= transverse relaxation time = time constant which determine the rate at which excited protons reach equilibrium/go out of phase
- measure of time taken for spinning protons to lose phase coherence among nuclei that are spinning perpendicular to main field
- -> both relaxation times (T1 + T2) depend on tissue type
MRI
- spatial encoding
= how we create an image, determine strength of signal at each frequency (= position)
- apply gradient in x direction: spins/precession frequency depend on their position along the gradient
- spatial info is then frequency-encoded; assign amount of signal to spatial locations - time domain signal sums all frequencies
- Fourier analysis: decomposes signal and shows amount of signal for each frequency
advantages of MRI over CT
- does not use ionising radiation (can be scanned multiple times)
- better spatial resolution: discrimination of individual gyri
- better discrimination between white and grey matter: enables early diagnosis of some pathologies
- can be adapted to functional imaging (fMRI)
functional imaging
= neural activity produces local physiological changes in that region, produce dynamic maps of the moment-to-moment activity
hemodynamic response methods
= hemodynamic response = when the activity of neurones increases, the blood supply to that region increases relatively to others (providing it with more glucose + oxygen)
- -> PET measures the change in blood flow directly (and the supply of different molecules to that region)
- -> fMRI measures the concentration of oxygen in the blood
PET
= positron emission tomography = measure local variations in cerebral blood flow that are correlated with mental activity –> increased blood flow to the brain regions that have heightened neural activity
- indirect measure
PET
- method
- radioactive substance is introduced/injected into the bloodstream (= tracer)
- possible to use radio-labelled neurotransmitters to investigate particular pathways + effects of drugs - radiation emitted from this tracer is monitored by the PET instrument
- radioactive isotopes within the injected substance decay by emitting a positron from their atomic nuclei
- positron collides with electron –> two photons (gamma rays) are created
- gamma rays move in opposite directions at speed of light - PET scanner (= gamma ray detector) determines where the collision took place
fMRI
= functional magnetic resonance imaging = measure the ratio of oxygenated to deoxygenated hemoglobin (= BOLD signal)
- similarity MRI: imaging focused on the magnetic properties of the deoxygenated form of hemoglobin
- allows to image time course of brain activity
- indirect measure: based on neuronal metabolic processes
fMRI
- physical properties
- when area becomes more activated/stimulated, the area consumes more energy (sources: oxygen, glucose)
- -> more active = more energy required (food) - neuronal tissue is receiving energy from oxygenated haemoglobin in the blood
- -> haemoglobin has the oxygen attached to it (in a bag) and brings you the food/energy - when neurones absorb the energy (=oxygen), the haemoglobin becomes deoxygenated
- -> you eat all the food, and haemoglobin leaves without oxygen attached to it - concentration of oxygenated haemoglobin in the blood increases in the active area, because the area requested it
- -> haemoglobin/blood delivers too much oxygenated haemoglobin (too much food)
fMRI
- BOLD
= blood oxygen-level-dependent contrast = paramagnetic properties of deoxygenated haemoglobin introduce distortions in the magnetic field –> distortions indicate the concentration of deoxygenated haemoglobin in the blood
- which indicate that energy has been consumed?
- -> if there is low concentration of deoxygenated haemoglobin in the blood, the area needed a lot of energy (= was active)
fMRI
- deoxygenated + oxygenated haemoglobin
- deoxygenated haemoglobin (Hb)
- is paramagnetic = distorts the magnetic field
- when area inactive, Hb concentration is high –> disturbs homogenous magnetic field of scanner –> fast dephasing –> low (f)MRI signal - oxygenated haemoglobin (HbO2)
- is demanded when neurone becomes active (energy) –> HbO2 concentration is high (less Hb) –> less distortion; more homogenous magnetic field –> dephasing takes longer –> stronger signal over time
- -> precessing frequencies differ
hemodynamic response function
= HRF = changes in the BOLD signal over time; way that the BOLD signal evolves over time in response to an increase in neural activity
- initial dip: neurones consume oxygen –> small rise in amount of deoxygenated haemoglobin (that was already in the region) –> reduction of BOLD signal
- will always limit our temporal resolution
- -> you are so hungry, that you aren’t hungry anymore - overcompensation: increased consumption of oxygen, blood flow to the region increases –> increase in blood flow is greater than the increased consumption = BOLD signal increases
- -> you eat - undershoot: blood flow and oxygen consumption dip before returning to their original levels –> temporal increase in deoxygenated haemoglobin (oversupply: area does not require that much energy anymore)
- may reflect relaxation of blood system
- -> you are shortly tired after eating a lot
- HRF relatively stable across sessions with same participant in same region; variable across different regions within same individual and more variable between individuals
- different HRF can be superimposed on each other
data analysis
- preprocessing
= needed to improve signal-to-noise ratio
preprocessing
- motion correction
= compensate for small head movements
- motion can cause voxels to locate differently
- after recording, before data analysis
a) detection: if we detect misplaced area and we know how much it changed, we can rotate it along x, y, z axes
b) correction: undoing detected motion; spatial interpolation is necessary
preprocessing
- stereotactic normalisation
= map regions of each individual brain onto a standard brain
- each brain divided into thousands of voxels which are given 3-dimensional spatial coordinates (x, y, z)
- -> Talairach coordinates (origin: anterior commissure)
- x (left-right) + y (front-back) + z (up-down) - every coordinate is then mapped onto the corresponding coordinate on the standard brain
- needed to compare between subjects
preprocessing
- temporal filtering
= remove slow (low-frequency) temporal drifts from data
- due to scanner-related noise
- they reduce statistical power and invalidate event-related averaging
- need to remove drifts without removing task effect
- -> frequency analysis (Fourier): filter out low frequencies at frequency domain and left without drifts (only task effects) at time domain
preprocessing
- spatial smoothing
= improve spatial and temporal signal-to-noise ratio; improve signal quality
- smoothing includes activity of neighbouring areas –> produce bigger activity fields
- average neighbouring areas and replace central voxel by weighted average of neighbourhood
- -> shape of Gausian function: middle voxel has highest weight; the further away, the smaller the influence on centre
- often reduces actual resolution because it disregards the activity of small regions
data analysis
- voxel-wise analysis
= voxels prefer one task or another (more active in one condition or another)
- shows where certain conditions differ from another
- compared to H0 distribution: want to surpass a certain threshold
- -> need to choose statistical threshold based on spatial smoothness
data interpretation
- activation in image is the difference between two conditions (Task A – Task B)
- -> regions can be activated/ deactivated
- BUT: Functional imaging may give us regions that are sufficient for the task rather than regions that are necessary (correlation rather than causation)
- -> active: doesn’t mean it’s essential for the task (could reflect strategy, attention, region receives input but isn’t responding to it (= inhibition))
- BOLD signal more sensitive to input rather than output
spatial resolution
= minimum resolvable distance between two closely spaced, simultaneously active foci; voxel that represents the minimum unit of brain tissue sampled in each image
- how good can we separate voxels
a) increasing voxel size = lowering resolution = increased amount of tissue detected as active
b) reducing size = increasing resolution = decreased sensitivity to BOLD effect but more spatially specific information + reduced susceptibility artifacts - more Tesla/ stronger magnetic fields
- -> resolution can be increased only at the expense of SNR + time
- limited by blood system
temporal resolution
= smallest detectable interval between two separate responses in the same/different activation focus
- discriminate processes in time
time between two excitation pulses (= time it takes to collect a brain volume composed of many slices)
- time between pulses that we use to switch the protons
- shorter TR –> less collected slices –> limited brain coverage
- temporal characteristics of HRF limit usefulness of very rapid image acquisition (no need to collect images at a TR lower than 1s)
triangular relationship
- research question determines which resolution you focus on
a) if spatial resolution is increased while keeping the temporal resolution fixed: amount of brain tissue sampled (no. of image slices) has to be reduced (increase spatial resolution + constant temporal resolution –> amount of brain tissue of interest decreases)
b) if spatial resolution is increased while keeping brain coverage: the temporal resolution has to be less, increasing experiment time (increase spatial resolution + constant brain coverage –> temporal resolution decreases)
combine spatial + temporal resolution
- jittering: use different delays between start of sampling of volume images relative to the start of the stimulus presentation (gaps in event-related)
a. if all images collected with same delay (= time-locked) all brain regions would be sampled at same time point at every ISI
- if one offsets the stimulus presentation time to image acquisition, the different time points would be sampled at each stimulus acquisition
- con: requires more trials + stable behavioural performance across trials
- advisable when: full brain coverage + temporal resolution needed but behavioural analysis + long scanning sections are not critical - parallel acquisition: uses difference in MR signal measured by coils, which depends on the proximity of the part of the body
a. reduces acquisition time by one specific factor set by experimenter
b. reduces amount of susceptibility artifacts, improving detection of signal from basal frontal + mesial temporal regions
