neuroimaging Flashcards
function of an MRI scanner?
developed for structural imaging
also used for functional brain imaging
1.5-7 tesla magnet
what is a BOLD signal?
Blood Oxygen Level Dependent signal
blood flows to active neurons in the brain to provide oxgen to fuel the cells
hemoglobin differs in how it responds to magnetic field, depending on whether it has a bound oxygen molecule (reduces signal if no oxygen bound)
MRI scanner detects these changes in the magnetic field
describe fMRIs?
functional imaging
doesn’t directly measure brain activation but blood flow
measures the magnetic properties of oxygenated which is going to active regions of brain vs. deoxygenated blood
what is the order of the steps to get from raw data to functional brain ‘activation’ maps
- design a task to be used in scanner (experimental and control condition)
- collect some data
- preprocess the data (clean it)
- analyse the data (whether getting results expected)
- interpret your results
what is meant by the BOLD signal as being ‘arbitrary’?
how to control for this?
it has no stable baseline : if you scan someone one day their singal may differ to the next day
so have to have both experimental and baseline condition
good baseline?
differs from the experimental condition only by the process of interest
e.g face processing
how long does it take from stimulus presentation to activation in brain area?
4-5 secs after stimulus presented
what is a block design?
disadvantages?
clumping all trials together in one block
- stimuli very predictable so alter strategies
- inflexible
- ecological validity problems
- can’t separate data by performance
what is an event related design?
advantages?
trials of different conditions are randomly intermixed rather than alternating
- can know which BOLD signal associated with which condition
- flexibility
- avoid practice events and predictability
- post hoc sorting
- temporal dynamics (how response changes over time)
what is meant by a volume?
how long to collect a single volume?
a scan of the brain from top to bottm
om
2-3 secs to collect single volume
what is a voxel?
and what are they used for?
a segment of the brain having been divided into cubes
so can easily reference specific parts of the brain
3x3x3 mm
what are the steps of preprocessing data?
and what does this mean
to ‘clean up the noise’ (non-task-related variables)
1) high pass filtering
2) motion correction
3) slice time correction
4) coregistration
5) normalisation
6) spatial smoothing
what is high pass filtering?
remove low frequency oscillations from data
what is motion correction?
correcting images of head movements while in scanner
what is slice time correction?
correcting for delay in different slices imaging in the brain
what is coregistration?
overlay functional images on structural image to see where in brain activation is occuring
what is normalisation?
account for size variability in subjects brains by warping each brain in standardised space (average brain)
in order to compare brains
what is spatial smoothing?
modify BOLD signal in a voxel according to the activation in a neighbouring voxel by using the Gaussian kernel function
as if a neuron fires, neurons close to it will fire as well so data fits this model
what is used to analyse data?
multiple regression
determine effect of IVs on DV (brain activation)
perform a contrast (t test) in each voxel
yellow indicates higher t values
then apply a threshold (p<0.05) to determine whether significant
what is the type 1 error issue?
how to correct?
with 100 independent t tests have 99.4% chance of type 1 error
so very important to correct for multiple comparisons (correct p value to fit data) BIG PROBLEM FOR REPLICATIONS AS UNCORRECTED RESEARCH (many occur by chance)
what are the 2 approaches to data analysis?
- whole brain analysis
2. region of interest analysis
what is whole brain analysis?
advantages?
disadvantages?
examine effects on a voxel by voxel basis across whole brain
advantage:
no need for prior hypothesis
disadvantages :
can lose spatial resolution and produce loads of areas that are difficult to interpret
what is region of interest analysis?
advantages?
disadvantages?
restrict analysis to a particular brain region
advantages:
avoids “laundry lists” of activated regions
hypothesis driven
avoids multiple comparisons problem so simple
generalisable
disadvantage:
easy to miss things going on elsewhere in the brain
not simple how to define regions of interest
limitations of fMRI?
- correlative (can’t say region activated is essential for that function) - need converging evidence
- temporal resolution is lower than EEG/ERP- need converging evidence
what does MRI measure?
measures the magnetic signal coming from hydrogen atoms (small dipole magnets - magnetic fields have axis and orientation)
MRI scanners have strong magnetic fields which makes hydrogen atoms in the body respond with a magnetic signal
