MRI

Question 1

Lamoor Frequency

Answer 1

When nuclei are placed in a magnetic field, a torque causes the moments to perform a precession motion similar to a spinning top. The larmor frequency is the precession frequency (measured in megahertz) of nuclei in a magnetic field, B₀.

Larmor frequency, f_L = gyromagnetic ratio, γ x magnetic field, B

The Larmor frequency for protons (hydrogen) at 1tesla is 42Mhz. (therefore gyromagnetic ratio for H is 42.6 MHz/T)

Question 2

Fixed magnetic Field

Answer 2

• Loops of wires with a large electrical current, kept at superconducting temperatures, which produces a large magnetic field (B0), in which a patient can be placed.
• A typical MR fixed magnetic field is 1-3 Tesla.
• B0 is also known as the longitudinal plane.
• As the coils are superconducting, the magnetic is always on.

When a patient is put in, the net magnetic moments of their Hydrogen protons align with the main magnetic field, producing ‘net magnetisation’ parallel to the main magnetic field.

Question 3

T1 / Longitudinal / spin-lattice relaxation time constant:

Answer 3

• The time it takes for the longitudinal relaxation vector to return to 63% of its original value after application of a 90degree RF pulse at the larmor frequency
• Its decay is exponential
• Is primarily caused by spin-lattice interactions
  
  • From dipole-dipole interactions between adjacent molecules.
  • Fat has a very short T1 time and relaxes rapidly. Water has a long T1 time and relaxes slowly. TR and TE are short - To maximize contrast, we pick a TR time intermediate to these two decay curves.
• Influenced by B_o and temperature (both increase T1), macromolecules, paramagnetic substances

Question 4

T2 / TRANSVERSE / SPIN-SPIN RELAXATION TIME CONSTANT

Answer 4

• The time it takes for the transverse magnetization to reduce to 37% of its maximial value after application of a 90RF at larmor frequency via dephasing
• Its decay is exponential
• Is primarily caused by spin-spin interactions (T2)
  
  • T2* - is dephasing from both spin-spin interactions and magnetic field inhomogenieity
  • Fat has a short T2 and dephases rapidly. Water has a long T2 and dephases slowly. TR and TE are short – to maximize contrast, we pick a TE time intermediate to these two decay curves
  • T2* is decay not just from spin-spin interactions, but also from chemical shift and magnetic field inhomogenieties as well.
• Influenced by Bo, temperature, macromolecules, paramagnetic substances (like gadolinium)

Question 5

Radiofreuency fields (coils):

Answer 5

• RF fields come in the form of rapidly changing electromagnetic fields generated by loops of wire applied perpendicular to the main main magnetic field
• These RF coils produce RF electromagnetic fields which interacts with the net nuclear magnetization vector
  
  • When Radiofrequency field pulse matches the larmor frequency, resonance happens
    
    • Produces a transverse component of magnetisation
      
      • The result of this is that the NMV moves away from B₀ alignment = flip angle
• Coils have a transmit and a receive function

Question 6

RESONANCE

Answer 6

• Resonance is the energy transmission that happens when something is subjected to the same frequency that it oscillates at
• Effect of resonance
  
  • Energy absorption – increased energy leads to more spin up nuclei, when the number is the same as spin-down then the NMV lies in the transverse plane
  • Phase coherence – magnetic moments move into phase with each other

Question 7

T1 Recovery (spin-lattice)

Answer 7

• Spin-lattice relaxation = t1 relaxation constant = longitudinal relaxation time
• The time it takes for the longitudinal relaxation vector to return to 63% of its original/maximal value after application of a 90RF pulse at the larmor frequency (is exponential and is graphed below)
  
  • Ie. the high-energy antiparallel spin-up vectors are slowly going back to low-energy spin-down vectors
• Is a result of spin lattice interactions (nuclei giving up their energy to the surrounding environments)
• T1 is made from short TR and short TE (see the graphs) in spin-echo sequences
  
  • Example: TR = 700ms and TE = 10-20ms)
• Fat is bright on T1 (and H20 dark)

Question 8

TR – Repetition time (the time from one 90RF pulse to the next)

Answer 8

• TR is the time from one 90RF pulse to the next
• Therefore, this determines how much longitudinal relaxation can occur
• Therefore, TR determines the amount of T1 recovery
• a long TR (see above) means lots of recovery and everything has returned to B₀, so there is poor differentiation between signals (no T1 weighting).
• Therefore, a T1 weighted image has a very short TR to emphasise T1 differences
  
  • To maximize contrast, we pick a short TR that is intermediate between fat and water to accentuate the differences

Question 9

T2 decay (spin-spin)

Answer 9

• Spin-spin relaxation = T2 decay
• T2 relaxation time constant is the time it takes for the transverse magnetization to reduce to 37% of its maximal value after application of a 90RF at larmor frequency (is exponential and graphed below)
• After 90RF pulse – the net mag. Vector (NMV) flips from longitudinal to transverse, and puts the transverse components in phase. After 90RF turned off, dephasing of transverse magnetization starts
  
  • T2 decay is primarily due to spin-spin interactions
  • T2* - Local variations in the magnetic field (local field inhomogeneity), as well as chemical shift and spin-spin interactions) cause the spins to rotate at variable rates and they begin to de-phase
    
    • This means the transverse magnetization vectors returns to original state
• T2 decay is caused by the magnetic fields of neighbouring nuclei interacting with each other (spin-spin)
  
  • T2 becomes shorter with increasing viscosity/decreasing molecular mobility, therefore T2 is shorter for solids than liquids
• To maximize T2 contrast, use a long TR and a long TE in spin-echo sequences

Question 10

TE = Time to Echo

Answer 10

• TE is the time-to-echo – the time from the RF pulse to the application of a signal peak in the receiver coil
• TE timing therefore determines how much transverse decay goes on
• Therefore, TE controls the amount of T2 relaxation
• A short TE (see above) will result in very little T2 decay, therefore there are minimal differences in signals between 2 tissues with different T2 times. This means a short TE has minimal T2 weighting.
• Therefore, a T2 weighted image has a long TE to maximize T2 differences.

Question 11

Proton density

Answer 11

• Is simply the number of protons per unit volume of the tissue
• Higher proton density = greater the signal available from that tissue = brighter on PD weighted imaging
• PD contrast is always present and depnds on the area being examined (and the patient)
• In proton density, you need to minimize BOTH T1 and T2 weighting.
  
  • Therefore, to minimize T1 you have a long TR
  • Therefore, to minimize T2 you have a short TE
• So, a long TR and a short TE give PD-weighted imaging
• Dephasing due to inhomogeneities (T2*) produces a rapid loss of coherenet transverse magnetisiation (and therefore signal), so that transverse magnetization is at zero before most tissues have had time to attain their T1 or T2 relaxation times.
• So, we need ways to compensate for T2* effects (dephasing from inhomogenities (+ normal T2 decay)).
• We use
  
  • Spin-echo pulse sequences
    
    • Spin-echo pulse sequences start with a 90RF pulse to flip NMV into transverse plane
    • In transverse plane, magnetization rapidly dephases due to T2* effects
    • Spin rephrasing is achieved by using a 180RF pulse at a time TE/2 to generate a SE at time TE
    • The SE sequences of 90RF and 180RF pulses is repeated after a repetition time (TR)
      
      • N.b. TR is the time between consecutinve 90RF pulses
  • Gradient fields
    
    • Slice selection (z-axis)
    • Phase encoding (y-axis)
    • Frequency encoding (x-axis)

Question 12

SPIN-ECHO PULSE SEQUENCES

Answer 12

• Spin-echo pulse sequences start with a 90RF pulse to flip the NMV into the transverse plane
• In the transverse plane, the NMV rapidly dephases due to T2* (field inhomogeneities, T2 decay)
• Spin rephrasing is then achieved by applying a second 180RF pulse at a time TE/2, which generates a spin echo (SE) at time to echo, TE
  
  • The time at which is 180RF pulse is given, TE/2, is also known as tau
• The Spin Echo (SE) sequence of events is then repeated after a time-to-repitition, TR (time to next 90RF pulse)
  
  • We still change TE and TR depending on whether we want to maximize T1 or T2 weighting
    
    • For T1 weighting, we want a short TR and TE
    • For T2 weighting, we want a long TR and TE
    • For PD weighting, we want a long TR (to minimize T1) and a short TE (to minimize T2)

Spin echo pulse sequences compensate for T2*!

Question 13

GRADIENT RECALL ECHO SEQUENCES:

Answer 13

• Gradient echo sequences (GRE) are the alternative to spin echo. They differ in 2 ways:
  
  • Utilization of gradient fields to generate transverse magnetization
  • Flip angles of less than 90 degrees
• The precessional frequency of protons varies with the strength of a magnetic field, as dictated by the larmor equation (f_L = gyromagnetic ratio x magnetic field strength, B)
• By using magnetic gradients, we have different strengths of the magnetic field at different points. This means that the precessional frequency of magnetic moments will be different, depending where in the magnetic field they are
  
  • As gradients cause nuclei precessional frequency to either increase or decrease, they can be used to either dephase or rephrase magnetic moments.
• • First, give the variable flip angle
    
    • First gradient applied – precessional frequencies at the lower end of the magnetic field slow down, whereas precessional frequencies of under the higher end of the gradient speed up. The magnetic moments of the nuclei are therefore dephased. T2* effects are happening.
    • A second gradient with an opposite ‘slope’ is then applied – the fast nuclei are now in a lower magnetic field and slow down, the slower nuclie are in a higher magnetic field and speed up. The gradient has therefore re-phased. A maximum signal is therefore induced in the receiver (the gradient echo)
      
      • Is similar to SPE in that it still uses TE and TR
      • But is different in that variable flip angles are used
        
        • A spoiler is the de-phasing gradient
        • A rewinder is the re-phasing gradient
• Variable flip angles are used
• Advantage of GRE - The TE can be much shorter than in spin echo imaging
  
  • Coz we are not using 180RF pulses, the minimum TE is shorter (therefore the TR can also be reduced)
    
    • Therefore, GRE images can be acquired quicker!
• Disadvantage of GRE - Gradients do NOT eliminate effects from magnetic field inhomogeneieties, unlike spin-echo imaging
  
  • Therefore T2* Imaging is only done with gradient echo imaging (GRE), coz spin-echo will remove T2*!

Question 14

SPATIAL LOCATION OF A PICTURE

Answer 14

• Magnetic gradients are used to code the spatial location of the MR signal
• Remember, the precessional frequency of protons varies with the strength of the magnetic field (Larmor equation)
• By using gradients (and gradient field coils) in the x/y/z planes, we can code the spatial location of an MR signal
• 3 magnetic field gradients are
  
  • Slice selection(always the z-axis)
    
    • done when a pulse is applied (and at TE when collect echo)
  • Phase encoding – short axis
    
    • after RF pulse, before echo
  • Frequency encoding - long axis
    
    • done when reading out
• Magnetic field gradient pulse induces non-uniformities in precessing phase and frequency between volume elements between selected slices.
  
  • Using image reconstruction methods such as the Fourier Transformation, we can determine which volume element gave rise to which signal strength based on the specific phase and frequency of that volume element

Question 15

• Slice selection (always the z-axis)

Answer 15

• when the gradient field is applied, the precessional frequency of nuclei located along its axis is changed based on where the protons lie within the magnetic field (and how strong it is in that place) – this is Larmor equation.
• A slice can therefore be selected by transmitting RF with a band of frequencies coinciding with the Larmor frequency in that particular slice (tuning fork analogy)

Question 16

• Phase encoding

Answer 16

• The phase encoding gradient is switched on after the excitation pulse.
• Prior to the phase-encoding gradient, all the moments are precessing in phase (due to the 90RF pulse). When the phase encoding gradient is applied, protons precess at frequencies determined by the larmor equation.
• Once this phase encoding gradient is switched off they return to precessing at the same frequency, but all protons at a given y-coordinate will be out-of-phase with all other protons in the slice – this lets us determine their exact location

Question 17

• Frequency encoding

Answer 17

• When the frequency encoding gradient is switched on, the magnetic field strength (and thus the precessional frequency of the signal along the axis of the gradient) is altered in a linear fashion.
• The gradient thus produces a frequency difference or signal shift along its axis, allowing its position to be located along the axis of the gradient according to frequency
• The steepness of the slop of the frequency encoding gradient determines the field of view

Question 18

Chemical shift

Answer 18

• The electron cloud of an atom screens the proton from an external magnetic field, and results in the nuclei precessing at slightly different larmor frequencies.
• Since larmor frequencies are field strength dependent, it is customary to quote the chemical shift as the fractional shift in larmor frequency in parts per million
  
  • E.g. chemical shift between fat and water is 3.3 ppm (buts its impact increases with field strength)
    
    • In MRI, this can be seen apparent displacement of fat/water boundary in adipose tissue, and causes artifactual void
• This only occurs along the frequency encoding axis
  
  • Another explanation: the external magnetic field (B0) causes the electron cloud surrounding the nucleus to induce a tiny electron current that in turn produces its own local magnetic field at the nucleus in the opposite direction to the external magnetic field
• Appearance
  
  • Chemical shift artefact causes a signal void between areas of fat and water
• Remedies
  
  • Scan with a low field strength magnet
  • Remove either fat or water signal by use of STIR/chemical/spectral pre-saturation
  • Broaden the receive bandwidth

Question 19

SAFETY IN MRI

Answer 19

• Missile effect – ferromagnetic objects may be pulled into the magnet (think brain clips)
• The magnet is always on – the fringe field (peripheral magnetic field) will interact with electronics
• Magnetic fields interacting with implant electrical circuity – especially cardiac pacemakers
• Induction of electrical currents in conductive materials (nerves) – can induce pain
  
  • In America, limit of 3T/s to prevent peripheral nerve stimulation
• Radiofrequency can cause biological damage via heating – ensure body increases do not exceed 1 degree. Be especially careful if have monitoring leads/electrodes on skin, as these act as antenna and can cause burns.
• MRI create a LOT of noise (65-120dB) – hearing protection is mandatory
• MR needs shielding – faraday cage (using copper) for protection from RF, and iron plates provide magnetic shielding
• Gadolinium – nephrotoxic
• Pregnancy – evidence of MR harming a foetus is limited

Question 20

• Factors affecting T1 relaxation time

Answer 20

• Type and mobility of nucleus
  
  • Small molecules like water have a broad range of motional frequencies à poor matching with Larmor frequency à low efficiency for relaxation
  • Medium sized molecules like fat have a narrower distribution with a comparatively large amplitude for relaxation at typical resonant frequencies à shorter T1 time
  • Large molecules tumble too slowly to be effective at relaxation à long T1
• Magnetic field strength
  
  • Efficiency for T1 relaxation decreases at higher frequencies
  • Greater field strength = longer T1
• • Increased temperature = longer T1
• Presence of macromolecules
  
  • Increases T1
• Presence of paramagnetic ions
  
  • Power relaxation mechanism
  • Unpaired electron creates a magnetic moment ~700x larger than proton à v. large fluctuating field
  • Example = gadolinium

Result = shortening of T1

Question 21

• Factors affecting T2 relaxation

Answer 21

• Field strength
  
  • Increased field strength = longer T2 decay
  • Much less so than T1
• • Increased temperature = longer T2 decay
• Mobility of nuclear species
  
  • Pure liquids have no static intrinsic magnetic fields à slow T2 decay à appears bright e.g. CSF on T2-weighted
• Presence of macromolecules
  
  • Large macromolecules have large intrinsic fields at low re-orientation rates à efficient T2 relaxation
  • Large macromolecules e.g. fat appear dark on T2-w sequences
• Presence of paramagnetic ions or molecules
  
  • Gadolinium = reduction in T2 relaxation time

Question 22

• Spin echo

Answer 22

• Mechanism
  
  • 90° excitation pulse followed by a 180° rephasing pulse followed by an echo
  • Rephasing eliminates the effect of magnetic field inhomogeneity (compensates for T2*)
• Uses
  
  • Gold standard for image quality for T1, T2 and PD images
• Advantages
  
  • Image quality
  • True T2 weighting
  • Versatile
  • Available on all systems
• Disadvantages
  
  • Long scan times

Question 23

• Gradient echo

Answer 23

• Mechanism
  
  • Uses a flip angle <90 degrees
  • RF excitation pulse is followed by negative dephasing gradient to produce intentional dephasing. Positive gradient then applied to produce rephrasing of spins and generate an echo
  • The magnitude and duration of the RF excitation pulse selected determines the flip angle
• Advantage
  
  • Allows for selection of very short TE values
  • Faster (Coz using variable flip angles)
• Disadvantage
  
  • Unable to recover dephasing due to static field inhomogeneities
  • Represents a complicated function of T2* rather than T2

Question 24

Fourier Tansform reconstruction

Answer 24

• 2D FT recon
  
  • Fourier transformation of frequency and phase (pseud-frequency) values from K-space generates a complex image with ‘real’ and ‘imaginary’ parts
• Artefacts from Fourier imaging
  
  • Gibbs artefact
    
    • Ringing of signal on sharp edges in the image
    • Due to inadequate k-space values to represent the detail
  • Phase wrap
• Speeding up reconstruction
  
  • Half fourier/Half NEX
    
    • Just over half of phase-encoding steps omitted
    • Other half of data estimated using a mathematical trick
  • Reduced matrix
    
    • Largest phase-encode data omitted
  • Rectangular FOV
    
    • FOV reduced to size of anatomy (redundant PE steps eliminated)

Question 25

MRI Artefact: **Magic angle**

Answer 25

At 55degrees, high signal on short TE sequences (i.e. T1)

Question 26

**MRI Artefact: Phase wrap-around**

Answer 26

Is from aliasing (i.e. from incomplete sampling) – increase read sampling rate in frequency encoding direction to overcome

Question 27

Flow artefact: Physiological movement

Answer 27

* Use fast scans * Increase NSA * Respiratory gating * ECG gating

Question 28

* Motion artefacts from flow

Answer 28

* Spin echo sequences – excited blood moves out of slice; echo cannot be generated * Gradient echo – excited blood moves out of slice but still creates a signal; Unlike spin echo, the gradients which create the echo are not slice selective, so the excited blood still contributes a signal even though it is no longer in the image slice. * Avoiding flow artefacts * Saturation bands – 90 degree pulse applied to tissue immediately before the slice in order to saturate the signal so blood cannot interfere with the image * Gradient moment nulling – extra gradient pulses inserted to null the phase of moving blood

Question 29

* Tissue heterogeneity and foreign bodies

Answer 29

* Chemical shift (discussed earlier) * Magnetic susceptibility * Partial volume artefact * Partial volume artefacts occur wherever a voxel contains a mixture of tissue types * Reduced by using thin slices

Question 30

Fourier transform and Nyquist sampling theorem

Answer 30

* Phase wrap around artefact * The phase wrap-around artefact arises whenever the anatomy continues outside the field of view (FOV). It causes images of the tissues just outside the FOV to be produced at the opposite edge of the scan in the phase-encode direction * N.B. For greater fidelity in signal conversion, the sampling rate should be at least twice the highest frequency within the signal (Nyquist rate) * * Saturate the signals just outside the FOV using saturation bands * Phase oversampling - increases the FOV in the phase-encode direction and also increases the number of phase-encode steps so that the pixel size remains the same

Question 31

* Gibbs’ artefact

Answer 31

* Series of lines in the MR image parallel to abrupt and intense changes in the object at this location, such as the CSF-spinal cord and the skull-brain interface * At high-contrast boundaries (jump discontinuity in mathematical terms) the Fourier transform corresponds to an infinite number of frequencies, and since sampling is finite the discrepancy appears in the image in the form of a series of lines * * Increasing the matrix size (i.e. sampling frequency for the frequency direction and number of phase encoding steps for the phase direction) * Use of smoothing filters (2-D Exponential filtering, Gegenbauer reconstruction etc.) * if fat is one of the boundaries, use of fat suppression

Question 32

Signal to noise ratio

Answer 32

* Signal to noise ratio * Definition: ratio of amplitude of MR signal to amplitude of background noise * Factors affecting SNR * Field strength * Higher field strength improves SNR as more spin up nuclei available to generate a signal * Proton density * More protons = better SNR * Low density = chest * High density = pelvis * Coil type and position * Small coils are better than large coils but cover a smaller area * Good compromise is the phased array coil that uses multiple small coils that provide good SNR and these are combined to create an image with good coverage * TR * Short TR reduces SNR (less magnetisation available) * TE * Short TE improves SNR (more magnetisation available) * Flip angle * Greater angle improves SNR (more magnetisation available) * NSA * Increased NSA improves SNR (but also significantly longer scan time) * Receive bandwidth (range of frequencies sampled during readout) * Smaller bandwidth reduces proportion of noise sampled relative to signal

Question 33

Spatial Resolution

Answer 33

* Definition: ability to distinguish between two points that are close together in the patient. Entirely controlled by the size of the voxel * Factors that affect voxel volume * Slice thickness * FOV * The matrix * Larger voxels have more signal which results in better SNR. However smaller voxels improve resolution as they increase likelihood that 2 points close together in the patient will be in separate voxels and therefore distinguishable from each other