Neuroimaging Flashcards
1
Q
What are the different brain imaging techniques?
A
- structural (take image of brain at one point in time): CT and MRI
- functional (take image of active brain at multiple points in time): PET and fMRI
2
Q
What is CT?
A
- moderately invasive
- inexpensive
- widely available
- low spatial
3
Q
What is MRI?
A
- non-invasive (use radio frequency fields)
- extremely high spatial resolution
- number one choice for structural brain imaging in neuroscience research
- extreme magnetic field so no metal
4
Q
What is PET?
A
- moderately invasive (radioactive tracer)
- measures indirect metabolic correlates of neural activity
- can measure synaptic transmission directly
- high spatial resolution
- extremely low temporal resolution
- extremely expensive
5
Q
What is fMRI?
A
- non-invasive
- measures indirect metabolic correlates of neural activity
- high spatial resolution
- low temporal precision as it measures processes
- moderately expensive
6
Q
What is the BOLD (blood oxygen level dependent) signal?
A
- when neurons become active, blood flows to part of the brain to provide oxygen to fuel the cells
- haemoglobin (iron-containing oxygen transporting protein present in blood) differs in how it responds to magnetic fields depending on if it’s bound to oxygen molecule
- MRI scanner detects these changes in the magnetic field
7
Q
How is raw data converted into functional brain ‘activation’ maps?
A
- design a task to be used in the scanner
- collect some data
- pre-process the data
- analyse the data
- interpret results
8
Q
What is the experimental design of fMRI?
A
- BOLD signal is arbitrary (has no stable baseline) so needs experimental and baseline condition
- good baseline is one that differs from experimental only by process of interest
9
Q
What are block designs and their limitations?
A
- BOLD signal is slow (peaks 4-5 seconds after stimulus, 16 seconds to return to baseline
- often group together lots of trials
- all fMRI experiments originally employed block designs
- limitations: highly predictable, inflexible for more complex tasks, ecological validity, can’t separate trials by performance
10
Q
What are event-related designs and their advantages?
A
- trials of different conditions are randomly intermixed and occur close together in time
- BOLD signal relating to different trial times can be disentangled, can see which one increases the BOLD signal
- advantages: allows for complex and novel experiments, flexible, can be randomized, post hoc sorting, can look at novelty and priming, and at temporal dynamics of response
11
Q
How is data collected?
A
- scanner collects data in slices
- 2 to 3 seconds to collect single volume
- to reference points in the brain the image is divided into cubes or voxels
12
Q
What are the preprocessing steps?
A
- high pass filtering: remove low frequency oscillations
- motion correction: correcting where it may have moved
- slice time correction: top slice is collected seconds before nose slice so are aligned temporarily
- coregistration: overlay any functional activations onto a structural image to see which parts are activated
- normalization: warp brains into standardized space
- spatial smoothing: model data according to this property of neurons
13
Q
What occurs in analysis?
A
- multiple regression used to determine effect of number of IVs on single DV
- for each voxel multiple regression is used to estimate how closely the BOLD signal correlates with timecourse of each condition
- then perform a contrast
14
Q
How are multiple comparisons corrected?
A
- brain images divided into up to 130,000 voxels so 130,000 individual t-tests
- with that many t-tests and alpha level of 0.05 it’s guaranteed to find some significant voxels by chance
- alpha level needs to be adjusted to correct for multiple comparisons
15
Q
What are the different approaches to data analysis?
A
- whole brain analysis: examine effects on a voxel-by-voxel basis across the whole brain
- region of interest analysis: restrict our analysis to a particular brain region
