Lecture 6 MEG Flashcards
Invention of MEG
Who first described ‘Josephson Junctions’- layers of superconductor material separated by an insulator (or other non-superconducting material) that convert magnetic field fluctuations to voltage? (1962)
Josephson
Invention of MEG
Who first recorded alpha oscillations using an induction coil? (1968)
Cohen
Invention of MEG
Who produced the first single SQUID recording?
Cohen
Wherever we have electrical current flow, a magnetic field is also generated at right angles (perpendicular). How can we remember the direction of this magnetic field in relation to the electrical current flow?
The right hand rule
What is the ‘B’ field?
Another term for the magnetic field
Describe the basis of the MEG signal
As we learned last week, EEG signals are due to electrical polarization in neural tissue from synchronous firing of many neurons
Wherever we have electrical current flow, a magnetic field is also generated.
In MEG we measure the magnetic flux density
Which way in the neuron does the primary electrical current flow?
Down the dendrite
In MEG we measure magnetic flux density. What units are used to measure this?
Teslas (T)
As the magnetic fields measured in the brain are very small, we usually measure in femto-teslas (fT). To what power is a fT in relation to a T (how many decimal places is the T shifted)?
T -15
1fT= 0.000000000000001
15 decimal places to the right (14 zeros then a 1)
Why is it important that MEG data is collected in magnetically shielded rooms?
Neural signals are very very weak
It is important to minimise external sources of noise
What are two key MEG technologies?
Traditional whole head MEG
Room temperature atomic magnetometer systems
Which of the two key MEG technologies uses SQUIDs (superconducting quantum interference devices) and requires liquid helium to cool to around 4K?
Traditional whole-head MEG
Room temperature atomic magnetometer systems
Traditional whole head MEG
Which of the two key MEG technologies uses OPMs (optically pumped magnetometers) and does no require liquid helium, so is wearable?
Traditional whole-head MEG
Room temperature atomic magnetometer systems
Room temperature atomic magnetometer systems
What are SQUIDs and how do they work?
Super Conducting Interference Devices
When some materials are cooled to around 4K they lose all electrical resistance, these are known as superconductors
A current in a superconducting loop will keep flowing forever
If you apply a magnetic field to a superconductor, it will produce a current
These devices pick up magnetic signals in the brain
Why can SQUIDs not be places right next to the head? How to SQUID systems get around this?
They are too cold!
Pick up coils are used to relay the signal from the head to the SQUID via a flux transformer (the flux transformer is the pickup coil and a smaller coil that sits next to the SQUID)
What is a pick up coil?
A wire that moves the signal away from the head to the SQUID
Give 3 examples of types of coil used for pickup coils
Magnetometers
Planar gradiometers
Axial gradiometers
Different types of coil can record signals at different depths and angles
What are some disadvantages of SQUID-based MEG systems?
Expensive to run (liquid helium is spenny!)
Very few manufacturers left
Participants must be seated
Not great for scanning children with a smaller head- too far away from pick up coils
OPM is a newer MEG technological development than SQUID-based systems. What does OPM stand for?
Optically pumped magnometers
How do OPMs work?
- Laser light is ‘pumped’ through a cell containing vaporised rubidium atoms
- Magnetic field fluctuations change the transparency of the gas, and this affects the transmissance of the laser
- Each sensor costs around $7-10k
- They are very small (~2.5cm long)
- Less sensitive than SQUIDs, but can get closer to the head so performance is comparable
What are some advantages of OPM-based MEG systems?
- Newer technology, the first installations are now taking place
- Operates at room temperature (no helium)
- Cheaper to install and operate
- Sensors are mounted in a cap or custom 3D-printed head-cast
- Proprietary technology to actively null magnetic fields with coils
- Participants can potentially move around, and possible to test children
The two technologies are relatively comparable in the signals they give out. How are these signals acquired in both technologies?
- Magnetic fields are produced by the brain
- These are detected by a sensor and converted to an analog electrical signal
- For SQUIDs this involves a superconductor
- For OPMs it involves a laser and a photodiode
- Then we sample and digitize this electrical signal, usually at a rate of 1000Hz or higher (one sample every 1ms)
- This gives MEG excellent temporal precision
(Data analysis in sensor space)
In EEG, we can look at responses at individual sensors, is this possible with MEG?
Yes
(Data analysis in sensor space)
How many of these different types of analysis are possible with MEG?
Event-locked evoked responses
Frequency and time/ frequency analyses
Steady-state methods
Plotting topographical activity across the head
All 4
(Data analysis in sensor space)
In MEG data analysis- we plot ? vs ?
Magnetic flux density vs time
(A plot showing all the sensor responses is called a butterfly plot)
(Data analysis in sensor space)
What are some differences between data analysis in EEG and data analysis in MEG?
- There are some differences, particularly in terms of polarity
- ERP components are prefixed with M (for magnetic), e.g. the M170 instead of the N170
Data analysis in source space
Alternatively to sensor space, we can resolve our observations into source space
We work out which locations in the brain are responsible for the signals we measure
This is a difficult (ill-posed) problem, but there are several well-established solutions, give 3 examples
- Dipole fitting
- Minimum norm
- Beamformers
Why is it much harder for source space to be estimated for EEG?
Because immediate tissues (skull, scalp etc.) that are transparent to magnetic fields act to diffuse the EEG signal, and so must be modelled. The spatial precision is lower
MEG optogenetics!
- Optogenetics allows selective activation at an exactly known location with high temporal precision
- Detect these signals with MEG
- SAM beamformer localizes the source to within 1mm3
- Also works for subcortical regions like the hippocampus
- Source reconstruction can be really good under optimal conditions
info card
What is dipole fitting
A solution to data analysis in source space
- Fit a single source that best explains the sensor activity
- An equivalent current dipole
- Dot- where it is, Line- direction
- Can’t get a picture of activity across brain- for a specific location
What is minimum norm?
A solution to data analysis in source space
Find a solution that best explains activity at every vertex on the cortical mesh
What is beamforming?
A solution to data analysis in source space
- At each vertex independently, calculate a weighted combination of sensor activity (a ‘virtual electrode’)
- Best spatial precision
Step by step of source localisation (see notes)
Give some examples of uses and benefits of MEG
- Better spatial resolution than EEG (and source reconstruction is easier)
- (Much) better temporal resolution than fMRI
- Can record oscillations and measure dynamic connectivity
- It is silent, so particularly good for auditory experiments
- Wearable OPM systems are more child-friendly than MRI, so can be used for developmental/pediatric work
- Clinical utility:
- Epilepsy – can localize the source of seizures prior to surgery
- Traumatic head injury – potential for use in diagnosis/monitoring
