W4L2 - fMRI BOLD
No description yet. This curated deck is ready for daily study.
Card preview (16)
Difference between oxygenated and deoxygenated blood
Oxygenated (Hb): Diamagentic - Enhance Signal Deoxygenated (dHb): Paramagnetic - Decrease Signal (introduces distortions/artefacts)
What happens to oxygen when there is neural activity
Neural activity = Glucose Metabolism = Increase blood oxygenation = Oversupply of oxygenated blood = Better BOLD
Who discovered fMRI
Seiji Ogawa discovered that large blood vessels = brighter areas = better signal. Changed blood-oxygen level, make use of the fact that all neurons need oxygen from blood
Now that fMRI can see enhanced neural activity, what does it do
These areas of enhanced activity can then be mapped onto a structural image of the brain
What do we use to map areas of enhanced activity into the structural image of the brain
Statistical Parametric Mapping. Using a General Linear Model, we fit to brain activity at each measurement point/voxel.
What are blobs in fMRI
Blobs are averages. They are quality fits into expectations. They are voxels. “Activation “blobs” are statistical effects in experiment”
How do we get the BOLD signal
• BOLD signal within a region is measured while participants engage in a cognitive task • There is repeated measurement for the whole brain due to noisy signal. • Differences in BOLD signal tell us something about whether a brain regions ‘is engaged’ in the task
What does the BOLD actually tell us about neural activity
The BOLD signal is an indirect measure of neural activity. Enhanced neural activity impacts (a) how much oxygen is in the blood (b) how fast the blood flow (c) how much oxygen can be extracted. But the complex interplay suggest that neural activity = oversupply of oxygen
How do we interpret differences in BOLD signal. 3 Things
• Substantial temporal lag (8 Seconds) between neural activity and the peak of the BOLD response • Substantial temporal lag (16 seconds) before reaching baseline again • Invalid to compare signals between different regions of the brain because the signal change is different (only valid to compare between experimental conditions)
What is the measured response across regions in BOLD
Heamodynamic Response Function (HRF): This HRF is modelled when our data is analysed (i.e. we know how the shape of a “typical” response looks like)
What neural processes drive the BOLD signal
Feedback processing within excitation-inhibition networks (EIN) account for “activity” measured by BOLD fMRI. EIN determine what the output of the microunit will be. (Net excitation/Net inhbition) - does not tell us driving process
What are the limitations of BOLD
• ) Blobs does not suggest modularity • ) Poor temporal resolution • ) Good, but not great spatial resolution • ) Multiple Comparisons Problem
Limitation BOLD #1
Blobs does not suggest modularity • We do not see the full networks involved (might be other connectivity and information flow) • fMRI might not always map the functional units that matter
Limitation BOLD #2
Poor Temporal Resolution HRF slow - difficult to image very fast processes. - 2 Seconds to measure brain once, cannot explore processes within the 2 seconds To solve this: Events need to be spaced out by temporal jitters to obtain independent estimates of HRFs belonging to different events»_space;> But makes task artifical
Limitation BOLD #3
Good, but not great spatial resolution The smallest measurement unit is a “voxel”, which is a 3D pixel (3x3x3mm). Do not learn processes within a voxel, but one voxel contains >100k neurons.
Limitation BOLD #4
Multiple Comparisons Problem: Too much false positives Run t-test for each voxel (total >50,000) not possible. Use bonferroni-correction (divide sig level by no. of test) > but that makes sig findings difficult