Lectures fMRI Data and Statistics
Lecture 1 – 5 February 2019
What is MRI?:
Magnetic Resonance Imaging. Magnetic because the signal is created by a magnetic field and
Imaging because an image is made of the signal. Resonance because of the magnetic pulses
that are given to the body that create the spinning in phase of the protons.
In MRI protons are that what the image is made of. These protons especially consists of
water and fat. These protons are spinning and thereby creating a net magnetic field. This net
magnetic field is caused because not all water molecules orient in the same direction.
50,999% is oriented in the one direction and 49,999% is oriented in the other direction. This
very little difference causes the net magnetic field, which causes the signal that MRI image is
based on (slide 3). An antenna is needed to create the stronger magnetisation that can
actually be measured, since the net signal of the brain activity by the MRI scan is so small.
Where is MRI used for?
Understanding brain anatomy by imaging different brain areas.
Depiction and exploration of neuropathology and diseases (for diagnoses).
Localizing brain activity (related to certain tasks).
White matter fibers (DTI; structural connectivity).
Neuronal networks (functional connectivity).
Measuring cerebral blood flow (activity).
Microstructure (axonal diameter, etc).
MRI in four rules:
Rotation frequency scales linearly with local magnetic field strength:
The stronger the magnetic field, the faster the rotation of protons. So, it depends on the
amount of Tesla of the MRI scanner how fast the molecules spin (Hz). This relationship is
linear. A problem is that the magnetic field of a MRI scanner is not homogeneous (especially
not when there is a subject inside the scanner). A subject is not magnetic which causes the
variability within the magnetic field. To compensate for this problem, the disruption is
measured before the start of the actual tests and the scanner tries to homogenise the field
again. The more tesla, the more vulnerable the magnetic field is for variations and
disruptions.
Only magnetization in transverse plane is measured:
The longitudinal magnetic signal is pushed down in the transverse
plane, so it can be measured.
Relaxation processes determine contrast:
Basic MR parameters of tissue determined by three parameters:
proton density, T1 and T2.
Different matters in the body/brain have different proton densities.
These different densities are differently encoded in longitudinal magnetization. The higher
the density (the more water), the more signal. Difference in proton density creates the
contrast in the MR signal.
There are two important timing parameters:
o T1: How fast does the magnetization regrow in the direction of the main magnetic
field (longitudinal magnetization)?
T1 measures how quickly a magnetic field regrows in the direction of the equilibrium
situation. Takes about a second.
T1 short: fast relaxation.
, T1 long: slow relaxation.
Both short and long T1 start with the same longitudinal magnetization and rotate
with the same amount of transverse magnetization. However, the long T 1 takes
longer to regrow compared to the short T 1. If you rotate this again, then the long T1
has a shorter transversal magnetization. This affects the amount of signal between
different tissues.
o T2: How fast do we lose our transverse magnetization?
T2 short: fast relaxation.
T2 long: slow relaxation.
Takes about 80 ms. Tissue and fluid behave differently in the MRI signal, which
causes the contrast between different tissues on the image.
For both short and long T2 the same amount of longitudinal and transverse
magnetization is present. Because some spines move faster/slower the signal
spreads out. This can cause the signals to cancel each other out (in short T 2), which
causes low signal. This is called dephasing. When T2 is long, the spines stay together,
which causes a coherence signal and almost nog signal loss. This causes a strong
signal.
The time between two pulses is the echo time.
The influence of echo time (TE) and repetition time (TR) on contrast:
With low TE and low TR there is hardly any contrast. The higher the TE and the higher
the TR, the more contrast.
, T1 effects: rapid input of RF pulses or short TR.
T2 effects: long TE. Because some molecules take longer to go back to baseline.
Spatial localization based on spatial variation of magnetic field strength: to create an
image, you have to know where the signal is coming from. This is measured by the help of
gradients.
By switching on and off of gradients (the faster, the more detailed the signal is) gives forces
on the whole scanner (this is the reason why the scanner makes sounds). It is switched on
and off very fast, but not at the fastest rate to make that the scanner is not too loud and
otherwise you would feel little currents through your muscles which causes little twitches
which is very annoying.
Slice selection:
How can different slices be selected and imaged? This is based on the principle that changes in
magnetic field strength will automatically lead to a change in the rotation frequency of the protons.
With RF pulses (radiofrequency waves) the protons are pushed from the longitudinal to the
transverse plane. This is a coherent change of direction, which means that the protons will spin with
the same Hz in the longitudinal plane as in the transverse plane. Only the spins with the same
frequency will be transversed in the transverse plane and measured. This causes a slice. The
frequency in the RF determines the location of the slice and the amount of frequency determines the
thickness of the slice.
Distortions: Why?
Distortions in the MRI signal can be caused by non-uniformity of the signal, slice overlap or noise and
RF interference.
Typical fMRI acquisition
20-35 slices
In plane resolution of 3x3 mm
Slice thickness 3-5 mm
Dynamic scan time approximately 1 sec (was 2-3 seconds)
Gradient echo imaging
Use of single shot EPI: this is a gradient echo technique (echo planer imaging). It takes a
single shot slice by slice and has fast switching gradients. It is loud and contains distortions.
fMRI
In a block design you have to do something for a certain period (e.g. 30 seconds), then stop for a
certain time and repeat this a couple of times (to create enough statistical evidence). The signal
chances that can be seen are rather small (2%). The block design isn’t used anymore, because they
are currently using event related imaging.
The signal that is observed in fMRI arises from the BOLD response (Blood Oxygenation Level
Dependent). The dominant effect of BOLD is the blood oxygenation, but also the blood flow and
Lecture 1 – 5 February 2019
What is MRI?:
Magnetic Resonance Imaging. Magnetic because the signal is created by a magnetic field and
Imaging because an image is made of the signal. Resonance because of the magnetic pulses
that are given to the body that create the spinning in phase of the protons.
In MRI protons are that what the image is made of. These protons especially consists of
water and fat. These protons are spinning and thereby creating a net magnetic field. This net
magnetic field is caused because not all water molecules orient in the same direction.
50,999% is oriented in the one direction and 49,999% is oriented in the other direction. This
very little difference causes the net magnetic field, which causes the signal that MRI image is
based on (slide 3). An antenna is needed to create the stronger magnetisation that can
actually be measured, since the net signal of the brain activity by the MRI scan is so small.
Where is MRI used for?
Understanding brain anatomy by imaging different brain areas.
Depiction and exploration of neuropathology and diseases (for diagnoses).
Localizing brain activity (related to certain tasks).
White matter fibers (DTI; structural connectivity).
Neuronal networks (functional connectivity).
Measuring cerebral blood flow (activity).
Microstructure (axonal diameter, etc).
MRI in four rules:
Rotation frequency scales linearly with local magnetic field strength:
The stronger the magnetic field, the faster the rotation of protons. So, it depends on the
amount of Tesla of the MRI scanner how fast the molecules spin (Hz). This relationship is
linear. A problem is that the magnetic field of a MRI scanner is not homogeneous (especially
not when there is a subject inside the scanner). A subject is not magnetic which causes the
variability within the magnetic field. To compensate for this problem, the disruption is
measured before the start of the actual tests and the scanner tries to homogenise the field
again. The more tesla, the more vulnerable the magnetic field is for variations and
disruptions.
Only magnetization in transverse plane is measured:
The longitudinal magnetic signal is pushed down in the transverse
plane, so it can be measured.
Relaxation processes determine contrast:
Basic MR parameters of tissue determined by three parameters:
proton density, T1 and T2.
Different matters in the body/brain have different proton densities.
These different densities are differently encoded in longitudinal magnetization. The higher
the density (the more water), the more signal. Difference in proton density creates the
contrast in the MR signal.
There are two important timing parameters:
o T1: How fast does the magnetization regrow in the direction of the main magnetic
field (longitudinal magnetization)?
T1 measures how quickly a magnetic field regrows in the direction of the equilibrium
situation. Takes about a second.
T1 short: fast relaxation.
, T1 long: slow relaxation.
Both short and long T1 start with the same longitudinal magnetization and rotate
with the same amount of transverse magnetization. However, the long T 1 takes
longer to regrow compared to the short T 1. If you rotate this again, then the long T1
has a shorter transversal magnetization. This affects the amount of signal between
different tissues.
o T2: How fast do we lose our transverse magnetization?
T2 short: fast relaxation.
T2 long: slow relaxation.
Takes about 80 ms. Tissue and fluid behave differently in the MRI signal, which
causes the contrast between different tissues on the image.
For both short and long T2 the same amount of longitudinal and transverse
magnetization is present. Because some spines move faster/slower the signal
spreads out. This can cause the signals to cancel each other out (in short T 2), which
causes low signal. This is called dephasing. When T2 is long, the spines stay together,
which causes a coherence signal and almost nog signal loss. This causes a strong
signal.
The time between two pulses is the echo time.
The influence of echo time (TE) and repetition time (TR) on contrast:
With low TE and low TR there is hardly any contrast. The higher the TE and the higher
the TR, the more contrast.
, T1 effects: rapid input of RF pulses or short TR.
T2 effects: long TE. Because some molecules take longer to go back to baseline.
Spatial localization based on spatial variation of magnetic field strength: to create an
image, you have to know where the signal is coming from. This is measured by the help of
gradients.
By switching on and off of gradients (the faster, the more detailed the signal is) gives forces
on the whole scanner (this is the reason why the scanner makes sounds). It is switched on
and off very fast, but not at the fastest rate to make that the scanner is not too loud and
otherwise you would feel little currents through your muscles which causes little twitches
which is very annoying.
Slice selection:
How can different slices be selected and imaged? This is based on the principle that changes in
magnetic field strength will automatically lead to a change in the rotation frequency of the protons.
With RF pulses (radiofrequency waves) the protons are pushed from the longitudinal to the
transverse plane. This is a coherent change of direction, which means that the protons will spin with
the same Hz in the longitudinal plane as in the transverse plane. Only the spins with the same
frequency will be transversed in the transverse plane and measured. This causes a slice. The
frequency in the RF determines the location of the slice and the amount of frequency determines the
thickness of the slice.
Distortions: Why?
Distortions in the MRI signal can be caused by non-uniformity of the signal, slice overlap or noise and
RF interference.
Typical fMRI acquisition
20-35 slices
In plane resolution of 3x3 mm
Slice thickness 3-5 mm
Dynamic scan time approximately 1 sec (was 2-3 seconds)
Gradient echo imaging
Use of single shot EPI: this is a gradient echo technique (echo planer imaging). It takes a
single shot slice by slice and has fast switching gradients. It is loud and contains distortions.
fMRI
In a block design you have to do something for a certain period (e.g. 30 seconds), then stop for a
certain time and repeat this a couple of times (to create enough statistical evidence). The signal
chances that can be seen are rather small (2%). The block design isn’t used anymore, because they
are currently using event related imaging.
The signal that is observed in fMRI arises from the BOLD response (Blood Oxygenation Level
Dependent). The dominant effect of BOLD is the blood oxygenation, but also the blood flow and
