Week 3 - fMRI Flashcards
1
Q
What is fMRI?
A
- fMRI –> recording –> non-invasive –> hemodynamic
- fMRI is the most popular cognitive neuroscience method accounting for 57%
- Cognitive methods have their own pros and cons with regard to spatial resolution, temporal resolution, invasiveness, cost, etc.:
- fMRI provides a nice balance between these properties (good spatial resolution and OK temporal resolution)
- fMRI has become the dominant method in the past two decades
- fMRI was made possible by the discovery of BOLD (Blood Oxygen Level Dependent) signal (Ogawa et al., 1990)
2
Q
What is spatial and temporal resolution?
A
- Temporal resolution - the accuracy with which one can measure when an event (e.g., neural response) occurs:
- fMRI takes 1-4 sec to get one 3-D brain image
- EEG: 1000 Hz or higher per second – can tell more accurately when something happens
- Spatial resolution - the accuracy with which one can measure where an event (e.g., neural response) is occurring:
- Pixel = 2D
- Voxel = 3D, fMRI is voxel
3
Q
What are some limitations of fMRI?
A
- fMRI is noisy (as loud as 125 decibels)
- Police siren = 120 decibels
- Always give ear plugs or noise cancelling headphones to participants
- Not child friendly
* You’re not supposed to move in an fMRI scanner - Limited ecological validity in some cases
* You can’t ask participants to engage in real life tasks while in the machine, unlike using EEG
4
Q
What are the two major categories of brain scanning?
A
- Structural imaging:
- Still picture of the brain or body part (e.g., CT, PET, MRI) – can take up to 5 mins to get a picture
- Can study brain structures and diagnose disease/injury
- Can investigate how regional brain volumes are linked to individual differences in a particular cognitive ability (e.g., London taxi drivers have larger hippocampus).
- Functional imaging:
- Dynamic maps of the moment-to-moment activity of the brain (e.g., fMRI, PET, EEG) – you get many images with this technique
- Can study cognitive/affective processes
5
Q
What is MRI (Magnetic Resonance Imaging)?
A
- MRI gives pictures of the structure of the brain
- Both structural and functional images are acquired using the same scanner
- An MRI scanner consists of an electromagnet with a very strong magnetic field (1.5-10.5 Tesla)
- Can’t go into an MRI scanner with magnets on you
- Typically, 3 Tesla for human research
Principles of MRI (how we get data):
- Most human tissues are water-based
- The single protons that are found in water molecules in the body (the hydrogen nuclei in H2O) have weak magnetic fields
- In a normal state, these fields are oriented randomly
- When placed in an MRI scanner (i.e., magnetic field), they will align with the MRI magnetic field (point in the same direction)
- The machine applies a brief radio frequency (RF) pulse, which disrupts the proton and forces it into a 90-degree realignment
- Once this RF pulse is turned off, the protons realign with the magnetic field, releasing electromagnetic energy along the way, which can be detected by the MRI machine.
- MRI is able to differentiate various tissues based on how quickly they release energy (return to the original state) after the RF pulse is turned off.
- We can scan anything as long as the water molecule exists within
6
Q
How does fMRI work?
A
- The brain consumes 20 percent of the body’s oxygen uptake
- Haemoglobin is a molecule in the blood that can transport oxygen
- Oxyhaemoglobin – combination of haemoglobin and oxygen
- When neurons consume oxygen (energy), they convert oxyhaemoglobin to deoxyhaemoglobin
- When neurons in a particular brain region are active, there are more oxyhaemoglobin in that region – because neurons in that region need more energy
- Deoxyhemoglobin has strong paramagnetic properties, which introduce distortions in the local magnetic field.
- This distortion affects (suppresses) the fMRI signal
- The strength of the signal depends on the concentration of deoxyhemoglobin in the blood (the more deoxyhemoglobin, the lower the fMRI signal)
- fMRI signal measured in this way is called BOLD (Blood Oxygen Level Dependent) signal (Ogawa et al., 1990)
- fMRI is only an indirect measure of neural activity.
- Instead, it measures the metabolic demands (oxygen consumption) of active neurons
7
Q
What is the haemodynamic response function?
A
- The way that the BOLD signal evolves over time in response to an increase in neural activity is called the hemodynamic response function (HRF)
- The HRF has three phases;
- Initial dip – onset of neural activity, e.g. when light is presented, causes neurons to consume more energy which leads to more deoxyhaemoglobin so the BOLD signal is decreased.
- Overcompensation – system tries to compensate for the lack of oxygen so provides more oxygen which causes the BOLD signal to increase.
- Undershoot
8
Q
How well does BOLD signal actually reflect activity of neurons?
A
- Pretty well
- Simultaneous intracortical recordings of neural signals (single cell recordings) and fMRI responses from monkey visual cortex (Loghthetis et al., 2001)
9
Q
What are the three stages involved in fMRI data pre-processing?
A
- Head motion correction:
- Spatial normalisation
- Spatial smoothing
10
Q
What is head motion correction?
A
- Head motion is the largest source of variance in fMRI data
- Head motion could increase false negative (noisy data) – might be something interesting happening but you are not able to find it due to head motion
- We might find spurious activations due to systematic head motion (i.e., false positive)
- Head motion correction is done during data analysis
- But, it’s important to minimise head motion during data collection:
- Limit subject head movement with padding (make your subject as comfortable as possible)
- Give explicit instructions to lie as still as possible
- Keep each scan short (<8min)
11
Q
What is spatial normalisation?
A
- Mapping each individual brain onto a standard reference brain
- There are individual differences in brain size/shape for people
- To compare the results of different individuals, we need to map each brain onto a standard reference brain.
12
Q
What is spatial smoothing?
A
- Redistributing brain activity from neighbouring voxels to enhance the signal-to-noise ratio (blurring reduces high frequency noise while retaining signal).
13
Q
What is reverse inference?
A
- Reverse inference - “Inferring the engagement of a particular mental process from the activation of a particular brain region”
- Example - Amygdala is known to be activated by anxiety
- Amygdala is activated when an individual views an image of a spider
- Therefore, the individual felt anxious when viewing a spider
- People rely on reverse inference because:
- Researchers (psychologists) want to know about mental processes rather than neural process – for example, researchers want to know motivation underlying a particular social behaviour (e.g., donation)
- Reverse inference is often considered problematic:
- Any single brain region is not functionally specific.
- E.g. Amygdala is known to respond to anxiety, but studies also reported that it also responds to fear, happiness, etc.
- Insula is even more functionally heterogeneous; fear, disgust, anxiety, happiness, love, bodily awareness of emotion, pain, etc.
- Jauk et al. (2017) – reported narcissists don’t like looking at themselves:
- “Viewing one’s own face (as compared to faces of friends and strangers) was accompanied by greater activation of the dorsal and ventral anterior cingulate cortex (ACC) in highly narcissistic men.“
- “These results suggest that highly narcissistic men experience greater negative affect or emotional conflict during self-relevant processing”
- However, Insula is highly functionally heterogeneous.
14
Q
What are brain lesion studies? describing pros and cons
A
- Human:
- Patients with naturally-occurring brain damage
- Create “virtual lesion” using TMS/tDCS (brain stimulation methods)
- Experimental animals:
- Surgically remove brain regions
- Neurochemical lesions: kill neurons using neurochemical
- Historically, human brain lesion studies have greatly advanced our understandings of the brain:
- Patient HM: hippocampus & memory
- Patients with acquired prosopagnosia: FFA (Fusiform Face Area) & face recognition
- Patient SM: amygdala & fear (Week 9)
- Pros:
- Allow causal inference – between brain legions and cognitive abilities
- Precision is possible through anatomically or chemically selected lesions (animal lesion study)
- Cons:
- Brain damage may result in a reorganization of the cognitive system
- A discrete brain lesion can disrupt the functioning of distant brain regions that are structurally intact (called diaschisis)
- Take long time to collect data (human lesion study)
- Invasive (concerns over animal welfare)
15
Q
How do we test the causal role of a particular part of the brain in a normal subject?
A
- Brain stimulation methods
17
Q
Describe TMS (transcranial magnetic stimulation) - stating the pros and cons
A
- First demonstrated by Baker et al. (1985)
- How it works:
- When a strong, rapid current is passed through a stimulating coil, a rapidly changing magnetic field is produced, which induces current into the brain, causing increases or decreases in neuronal excitability (changes brain activity and function)
- Pros:
- Allow causal inference
- Temporary with no lasting damage (withinsubject design is possible)
- Cons:
- Researchers still unclear on how it works, exactly
- The effects of TMS are focal (1cm2, or so), but can only stimulate surface of the brain
- Expensive (~£30,000) than tDCS
