Briefly explain the broad mechanism of how fMRI works.
fMRI detects the difference between the organised spinning in oxygenated blood and disorganised spinning in deoxygenated blood.
Action potentials require oxygen and energy in the form of ATP (through glycolysis).
When oxygenated blood displaces deoxygenated blood in an area, the difference in proton spinning is detected.
The flow of blood to the area is called the Haemodynamic response (HDR)
So it measures neural activity indirectly.
What is a BOLD signal?
Blood oxygenation level dependent signal
Neurometabolic signal that has been shown to correlate with neural activity.
Recent work by __ has shown that BOLD signals correlate well with __ __ __ (__) spikes, but correlates better with what? What does this show?
Multi-Unit Activity (MUA)
Local field potentials (LFP)
LFPs are slower signals that reflects post-synaptic activity.
We can thus only say that the areas are involved in a task, not that the area is crucial. (ASK!!)
What do BOLD signals also reflect?
How active the neurons are
and how many neurons are active
Therefore, manipulations that change the activity of many neurons a little show larger activation differences than manipulations that change the activation of a few neurons a lot.
List the advantages of using fMRI.
2. Can scan whilst participant is performing a task: Investigating links between brain and behaviour
3. Whole brain coverage: Understanding broad networks and connections between different brain regions
4. BOLD signal correlates well with neural activity like LFPs
[Explicit testing of models derived from other approaches (eg. neurophysiology)
Informing and constraining theories of cognition
Understanding healthy development (non-invasive, longitudinal studies)
Whole brain coverage that can direct studies using other methodologies (eg. neurophysiology)]
What are the disadvantages of fMRI?
1. Low spatial resolution
2. Low temporal resolution(cannot understand components of cognitive processes)
3. Indirect measure of neural activity
4. Correlation, not causal relationship between BOLD and behaviour
5. Expensive! Hard to get good sample size.
List the 5 brain imaging methods that are complementary to fMRI.
Diffusion tensor imaging (DTI)
Transcranial magnetic stimulation (TMS)
What is diffusion tensor imaging?
Measures anatomical connectivity between brain regions - location and orientation of white matter tracts
Water diffuses easiest when there are no restrictions - thus it diffuses faster along length of fibre, not the width.
What is electroencephalogram (EEG)? (Also, on the side, provide the 2 types of potentials please Ron?)
Measures electrical potentials of many neurons associated with postsynaptic activity,
through electrodes placed on the surface of the scalp
Event-related potentials - as a result of cognitive event
Evokes potentials - potentials in response to sensory stimulus
Give the advantages and disadvantages of EEG.
1. High temporal resolution - it can investigate different components of cognition
2. Allows study of temporal interaction between different cortical regions
1. Low spatial resolution - which neural population is responsible for the behaviour?
2. Electrodes on scalp - not a great indication of neural activity in the inner brain regions
3. Again, like fMRI, not causal relationship
What is magnetoencephalography (MEG)?
Measures very small changes in magnetic fields caused by electrical activity of neurons using superconducting coils.
What are the advantages and disadvantages of Magnetoencephalography (MEG)?
1. Higher spatial resolution than EEG (unimpeded by resistence of skull)
3. High temporal resolution
1. Correlational approach - not causal
2. Some problems of determining which neural populations are related to behaviour
3. Can only detect electrical current tangential to the surface of the skull
Why is studying brain lesions alongside fMRI good?
Process evoking fMRI activity in brain region and is disrupted when the region is lesioned, then it can be concluded that the process relies on this region.
Especially useful for studying large scale systems
plasticity and recovery of function
What are some of the problems of using brain lesions?
Large and extensive rather than prescribed lesions across many functional brain regions
Large variability across patients
Damage to white matter pathways will impair functions supported by distant regions that are not damaged but become disconnected.
What is transcranial magnetic stimulation (TMS)?
Temporary stimulation of a brain region
by using an electromagnetic coil placed close to the scalp
TMS allows focal 'virtual lesions'
What is TMS used for?
Finding causal relationship between region and behaviour.
Determining whether regions activated in fMRI studies are essential for task performance
Potential treatment for disorders like depression by increasing levels of dopamine in frontal cortex
What are the two main types of fMRI experimental design?
Block design: Trials from different conditions are grouped into different blocks. Within one block, only one condition is presented. Thus the increase in fMRI signal in response to the stimulus is additive.
Event-related design: Trials from different conditions are randomized across the whole experiment. eg. slow ER design, rapid ER design
What is the difference between detection and estimation?
Detection is the measure of the ability to detect activity in certain voxels in response to a certain experimental manipulation.
Estimation is the measure of the ability to detect the time course within an active voxel in response to experimental manipulation.
Pros and cons of block designs.
1. High detection power (because signal is accumulative)
2. Ideal when network of areas involved in task is unknown and need to search the whole brain
1. Participants get used to the same stimulus and gets bored/brain may adapt, causing weaker signals.
2. Low estimation power (so time course of voxel activity is unclear
3. Cannot separate fMRI signal for single events (cannot tell difference between stimulus familiar/unfamiliar etc - just where it is active)
Pros and cons of event-related designs?
1. Higher estimation power - can detect signal associated with individual types of events (eg. familiar vs unfamiliar stimulus)
2. Ideal when area involved in task is already identified and you need to explore function of a brain region in more detail.
1. Reduced detection power - fMRI signals are weaker (need to collect x3 data as block design)
What are the two main types of fMRI data analysis?
Voxelwise Analysis: Search whole brain for voxels that show significantly higher responses to a stimulus/task compared to control condition.
Requires no prior hypotheses.
Region of Interest Analysis (ROI): fMRI signals of voxels only in this region are studied.
(can be functional or anatomical ROI)
Pros and cons of Voxelwise Analysis?
Pros: requires no prior hypothesis
covers the whole brain - nothing is missed
considers whole-brain networks
Cons: crompromised spatial resolution due to intersubject averaging.
Pros and cons of Region of Interest Analysis (ROI)?
Pros: Higher statistical power
Avoids problems with functional mislocation due to intersubject averaging.
Region much smaller - higher spatial resolution, reduces multiple comparisons problems
Cons: Neglects other areas which may play a fundamental role in task of interest
What are 4 sources of noise from fMRI? And how can they be corrected for?
1. Partial volume effects
Larger voxels provide higher signal-to-noise ratio because more gray matter (synapses & dendrites)
Smaller voxels and those with more white matter need to be scanned at a higher resolution with more advanced coils/higher field scanners to solve this problem.2. Physiological noise (eg.Head movements, heard beating, fidgeting)
Results in activations shifting over time, making it difficult to align brain images across the scan.
Computer algorithms can correct for some image shifts due to head movement. However, some distortions are hard to correct for.
Best prevention would be to give clear guidance not to move head
Large vessels line sulci and produce BOLD activation further from the true site of activation making it hard to tell where the activity arises from. Esp. problematic when comparing between age groups or populations that may differ in their vasculature.
4. System noise
Difference in brain voltage can sometimes reduce the sensitivity of the receiver coils.
[In the end though, the best way to prevent noise is to make it a good experimental design - reduce fidgeting or random thoughts etc]
To integrate results across subjects, what can you do?
Use brain atlas - adjust all brains to the standardised brain atlas
Look for key landmarks (such as the central sulcus etc)
List some of the ways to avoid confounds in fMRI studies?
1. Choose the right control condition: rest period blank is used for baseline but it is not a good control condition.
This is because any stimulus condition differs from the blank period in many respects that are not specific to the stimulus or task (eg. general arousal, attention etc)
Therefore, we need to contrast activity related to several stimulus conditions or tasks against each other.
Also, when comparing between tasks, difficulty should be adjusted so that the fMRI activations do not simply reflect task difficulty.
2. Avoid statistical errors:
Occurs when statistical tests perfomed are not independent from the means used to select the brain region
(eg. if you identify a region of voxels by contrasting activation between two conditions - AvsB - you cannot perform post-hoc comparisons using fMRI signals from the same voxels to infer that this region responds higher to A than B.
Use independent data (split-half reliability test) to identify brain region then evaluate fMRI reponses in this region to behaviour
Or use Leave-one-subject-out (LOSO) approach
What is structural magnetic resonance imaging? (sMRI)
Measures the structural/anatomical aspect of the brain
Water goes through different environments in the brain, and this is detected
Can locate brain damage or abnormalities (such as tumours) or we can quantify different aspects of the brain (cortical thickness, grey matter volume etc).
Why might meta-analysis be especially important when considering fMRI studies?
fMRI is expensive, noisy and often work with very small sample sizes
Therefore meta-analysis (a way of combining lots of studies to analyse what is common about them) is particularly important for psychological brain imagining studies