MRI Flashcards
Larmor
Precession frequency = ?
frequency = w = gamma x field strength
gamma = proportionality constant = gyromagnetic ratio
T1 term
spin lattice relaxation
T1 time
Time at which longitudinal magnetization is 63% of its final value
Greater field strength = longer T1
T2 fancy term
spin spin relaxation
T2 time
time at which signal has decayed to 37 % of its original value
what causes T2 decay (loss of transverse mag)
external field inhomogeneity
inhomogeneities in local field, within tissues. This is why pure things (water) take longer to decay and are therefore bright!
which is faster T2 or T2*
T2* decays faster
Tissue spin interaction + field inhomogeneity
TR =
time between initiation of two successive RF pulses
T2 or T2* decay (downslope)
180 degree RF
FID free induction decay
NO FID with a 180 pulse, just inverts longitudinal mag
otherwise all flip angles cause them
180 RF, T2 and T2*
after a 90 produces transverse mag, as T2 decay is happening, hit with a 180 degree RF pulse
this second, 180 pulse 1 clears out inhomos turning T2* into T2 and creates an echo
180 RF generates what?
turns T2* to T2 and generates echo
echo is some time after the 180 where transverse mag signal has refocused and peaks again in uniformity, then trails off again
When to deliver 180 RF
180 RF given at 1/2 time to echo
1/2 TE = 180 RF
Rough long and short TR
Long TR = > 2000ms
Short TR = 250-700 ms
long and short TE
Long TE = >60ms
Short TE = 10-25ms
Proton density times
MINIMIZE BOTH T1 and T2 effects
Long TR and Short TE
K space trivia
what’s at the center
what’s at the periphery
Center = information about gross form and contrast
Periphery = information about spatial resolution
Spatial encoding
slice select
encode spatial information along rows
encode spatial information along columns
localizing gradients
identical properties, applied at different times and different directions
Selecting desired slice
SSG
placed perpendicular to desired slice plane
Selective pulse
on top of a slice select gradient
RF pulse applied at same frequency as protons in slice being sampled, only protons in this plane will be affected
effect of 180 RF on gradient in spin echo
180 RF pulse applied after the 90 fucks up the field, so before and after the 180RF, you place two identical gradients to cancel each other out and correct for errors
gradients on either side of 180 RF
phase encoding
encoding spatial info in vertical direction
phase encode gradient causes protons in same row perpendicular to the gradient to have same phase. All protons will have same frequency at this point
phase vs frequency encoding
phase encoding is much longer. This is why it’s done on the thinner portion
DURATION of a 2d imaging sequence
phase or frequency encoding
DEPENDS ON PHASE ENCODING
Duration = TR x Phase encode steps x number of excitations
Horizontal = frequency encoding
frequency encoding
results in a column of protons which have identical frequencies
APPLIED at SAME TIME AS READOUT
TR =
time BETWEEN 90 RF’s
When is SSG applied?
at same time as 90 degree RF’s
Modifications
Table time
Time = TR x Phase matrix x # of excitations
Nex = number of times each set of phase encoding steps is repeated
Modifications
when does time not follow normal equation?
Fast spin echo
Acquisition time is proportional to 1/echo train length
3D scan time?
TR x phase matrix x NEX x #Slices
Modifications
Spatial res
primary factor
voxel size
Voxel = ?
slice thickness x (fov phase/matrix size phase x fov read/matrix size read)
Modifications - spatial res
FOV
SMALLER FOV = better spatial res
Modifications - Spatial res
Matrix size
Larger matrix = smaller pixels, better res
pixel = FOV/Matrix
Modifications - Spatial res
Gradient
higher amplitude, more intense gradient = better spatial res
Modifications - Spatial res
Slice thickness
how to change
thicker slice = increased transmit RF pulse or decrease (less steep) slice selection gradient
THINNER SLICE = BETTER SPATIAL RES
SUMMARY FOR BETTER SPATIAL RES
4 things
Small FOV
Big matrix
THIN slices (steep/large select gradient, thin transmit RF bandwidth)
Small voxel
Modifications - Signal to Noise
Voxel
OPPOSITE OF SPATIAL RES
BIGGER VOXEL = BETTER SNR
Modifications - SNR
voxel factors
BIGGER BETTER SNR
LARGER FOV
SMALLER MATRIX
THICKER SLICES (increased RF pulse, decreased SSG)
Modifications - SNR
Field strength
Stronger field = BETTER SNR
Modifications - SNR
coils
Smaller, surface coils improve signal
Modifications - SNR
excitations per slice (number of averages)
trade - off?
More excitations = more signal = BETTER SNR
INCREASED imaging time
Modifications - SNR
Receiver bandwidth
Fat bandwidth, more noise picked up (noise constant in time), lower SNR
NARROW BANDWIDTH = BETTER SNR
SNR summary
BETTER, 9 things…
Stronger magnet
Long TR
Short TE
BIG FOV
SMALL Matrix
THICK slices (weaker gradient, thick transmit BW)
MORE NEX
SMALL BANDWIDTH
APPROPRIATE COILS
Receiver vs transmit bandwidth
Receiver bandwidth, bigger = more noise = WORSE SNR
Transmit bandwidth, bigger = THICKER SLICE = BETTER SNR
PD and SNR
Good way to remember
LONG TR
SHORT TE
BOTH GOOD WAYS TO IMPROVE SNR
Tradeoffs
Field strength
Better SNR
INCREASED T1 time and therefore longer acquisition time via TR
more fielf, more Signal, longer time
Tradeoffs
NEX
NEX directly related to signal (better SNR) but only by factor of root-2
acquisition time goes up linearly
Spin echo basics
90 - PE, FE - 180 - echo, FE - 90
SSG with 90
180 at 1/2 TE, flanked by cancelling SSG, produces echo
FSE basics
?echo train length
Idea is to reduce TR (major contributor to length)
MULTIPLE 180’S
TR = time between 90’s with bunch of 180’s between
Echo train length = echoes in same TR
Fat signal and Fast spin echo
Fast repetition of 180 pulses cause T2 of fat to lengthen
J coupling
Acquisition time in FSE
approx proportional to 1/ETL
Inversion recovery
examples and times
START WITH 180
wait for relaxation time of thing you want to null (TI), hit with 90, this tissue gives no signal
STIR - 120-160 ms
FLAIR - 2000 ms
STIR- best fat supp sequence for metal or field inhomos
NO Gd with STIR (Gd has similar TI to fat)
IR tradeoff = increased acquisition time (longer TR)
GRE
No 180, flip angle less than 90
(no 180 means T2* not T2, so more susceptibility)
Lower SAR (less heat)
GRE
echo in GRE called?
Effect of shortened TR on transverse mag
Spoiled vs Refocused GRE
Echo in GRE = ‘Field echo’
Shortened TR means permanent residual transverse mag
Spoiled (incoherent) GRE = gradients or RF pulses used to get rid of transverse mag
Refocused (coherent) GRE = rewind gradient. example = SSFP
refocused = T2* weighted
Echo Planar imaging
aka
Single shot
single RF pulse
can be done with spin echo OR GRE
FASTEST SEQUENCE
turning the phase and freq encoding gradients on and off rapidly, very fast filling of K space
EPI artifacts
VERY susceptible to susceptibility (improved with segmented sequences)
Ghosting (gradient imperfections fuck with spatial encoding)
Chemical shift - narrow readout bandwidth used (fat bandwidth improves chemical shift)
DWI
Two gradients
base sequence
NON-moving molecules get hit twice, dephased then rephased —-> HIGH signal
MOVING molecules Hit once, move, second gradient misses original protons, LOW SIGNAL
B factor
Higher B factor = greater diffusion weighting
B Zero
No contribution from diffusion, usually long TR and TE, B zero basically a T2
fMRI
depends on ?
blood flow results in local reduction of deoxyhem
deoxyhem is paramagnetic, alters T2*
fMRI depends on T2*
2D TOF MRA
GRE saturation pulse to null venous or arterial flow
SMALL VOXEL SIZE
3D TOF MRA
higher SNR than 2D, good for high flow like COW
shorter imaging time
Phase contrast MRA
bipolar gradients to create contrast from flow
High VENC arterial, low venous
QUANTITATIVE IMAGE - can measure mean blood flow velocity and direction
Fat sat technique that’s not STIR
Selective pulse
Instead of using inversion time, works based on resonance diff bt protons in different microenvironments
PRECESSION FREQUENCY DIFFERENCE, more noteable with stronger fields
FAT SAT example- narrow RF pulse dropped only at resonance of fat, followed by SPOILER GRADIENT
ONLY dephases protons primed by initial narrow RF pulse (fat), so they can’t contribute to signal
Proceed as normal first RF pulse delivered immediately after spoiler
Out of phase timing
1.5T magnet
out 2.2 msec
in 4.4
3T magnet
out 1.1 sec
in 2.2 sec
Does it matter to miss 2.2 msec and start with in-phase?
Yes. IRON liver
Iron related T2* effect gets worse, darker, with time and second image will be darker with Fe present
Done correctly
Iron liver bright on 2.2 msec out of phase, darker on 4.4 msec in phase
Done wrong
Iron liver brighter on 4.4 sec in phase, darker on 6.6 out of phase, could wrongly assume it’s 2/2 fat
Type 1 chemical shift (fat) artifact
looks like
which sequences
Bright rim on one side, dark on the other
Spin echo or GRE
Type 2 chemical shift (fat) artifact
Looks like
what sequence?
Black boundary, India Ink
Type 2 goes all the way around
fat-water interface on GRE if a voxel is 50/50 fat water, signals cancel
Spin echo will eliminate
Type 2 only with GRE
Type 1 better/worse
3 things
Field strength - stronger = more chem shift
Gradient strength - increased gradient = less chem shift
Readout Bandwidth - Fatter = less chem shift
MRI contrast
2 main types
positive agents - shorten T1
negative agents - magnetic inhomog from susceptibility, T2 shortening (darker)
Gad
charge
chelated to
+3
DTPA chelated
How does Gd work
Gd has seven unpaired electrons, causes LOCAL augmentation of magnetic field
Gd and T2 effect
at most concentrations, T1 effect dominates (bright)
High concentrations (bladder), T2 takes over (DARK)
pseudolayer in bladder = layering dark in dependent bladder where concentration is high enough for T2 to dominate
MRI artifacts - image process
Aliasing
cause
direction
fix
wrapping of anatomy from under sampled portions
PHASE ENCODE direction
increase FOV or
Change PE direction
MRI artifacts - chemical shift
direction
types
FREQUENCY
type 1 bright - dark
type 2 dark all the way around
MRI artifacts - image process
Truncation/Gibbs
cause
look
direction
fix/penalty for fixing
K space fourier’d to a block of data, ripples occur at abrupt tissue interfaces with appearance of lines (csf/cord –> fake syrinx) “limited sampling of free induction decay”
More commonly in phase encoding direction (usually smaller PE matrix than FE)
Better = more matrix (decreasing bandwidth, pixel size, less FOV)
fixing means increased time and worse SNR
MRI artifacts - image process
partial volume
same as CT, different signal intensities overlapping in same voxel —> intermediate signal
fix with smaller pixels
MRI artifacts - Patient related
Motion
created differences between FE and PE
mainly seen in PE
MRI artifacts - patient related
Flow
SE vs GRE
Ghosting in PE direction
GRE more susceptible
SE flow is DARK (GETS HIT BY 90 THEN MISSED BY 180, NO ECHO)
GRE flow is BRIGHT
fix = apply a sat band (90 followed by crusher gradient)
MRI artifacts - Patient related
Magic angle
sequence?
MSK seen with tendons
SHORT TE SEQUENCES (T1, PD, GRE)
focus forms an angle of 55 degrees with main field
NOT seen on T2 (long TE)
MRI artifacts - RF related
Cross Talk
Overlap in RF and FT pulses, excitation of neighboring section more than once in a single repetition
Leads to partial separation and LOWER SIGNAL
3D images not susceptible to this
MRI artifacts - Zipper
cause
look
Stray RF pulses, inadequate shielding
zipper, 1 to 2 pixels in width running across, typically PE
MRI artifacts - External field
Inhomo fat supp
local inhomo’s allow certain areas of fat to resist suppression, mimic edema
fix = STIR, especially with metal
MRI artifacts - magnetic susceptibility
dia, para, ferromagnetic
most and least severe in which seqs
Susceptibility artifact due to metallic artifact can be reduced by increasing bandwidth, orienting the frequency encoding direction parallel to the long axis of the metal, increasing the field of view, decreasing the voxel size, and maximizing the echo train length. Decreasing slice thickness and echo time will also help.
Most severe with GRE
Less with SE (180 RF minimizes T2*)
T2* worsening with time (Fe liver getting darker on in phase)
also seen at tissue interfaces (bone and muscle, bone and air)
fixes - Use SE, swap directions, align metal with field, wider bandwidth
Shimming
goal
Types
Improve field homogeneity
Passive - phantom scanned and shims plates adjusted until field is homogenous. DONE AT INSTALL
Active - using electromag coil, can be done after each patient (or sequence)
MRI artifacts - Gradient related
Eddy Currents
cause
look
sequences
fix
Generated when gradients are turned on and off rapidly
occur in magnet, cables, wires or patient
looks like distortion, shift/shear
WORST WITH DWI (smeared frontal lobes)
fix by optimizing sequence of gradient pulses
MRI artifacts - Errors in data
Dielectric/standing waves
cause
look
fix
reduction of wavelength bc of dielectric constant..?
local eddy currents
WORSE with stronger magnet, large bellies with ascites
Dark signal in central abdomen over LEFT LOBE OF LIVER
fix with dielectric pads bt patient and anterior body coil
parallel RF transmission, RF pulses from a set of coils with independent RF pulse (longer pulse)
MRI artifacts - errors in data
Crisscross/herringbone
look
cause fix
Obliquely oriented stripes throughout
data processing or recon
Fix by reconstructing again
Cardiac MRI
Bright blood - type of sequence
Gradient
Cardiac MRI
Dark blood - sequence?
Spin echo
Double inversion recovery
Inversion recovery
normal time
around 330msec
nulled myocardium like flair or stir
Breast MRI
features of implant rupture sequence
fat and water saturated sequence (silicon will be bright)
Breast MRI
common artifact
chemical shift
fix = increase bandwidth
Breast MRI
Directions
Phase run side to side so pulsation stays out of tits
Breast MRI
Signal flair
breast to close to coil element, won’t fat sat out correctly
reposition patient
Quench
venting cooling liquid helium out of the room
resistance rapidly increases with temperature and magnet fails
When to quench
someone pinned by something metallic
Life threatening fire
Code in the scanner?
not a reason to quench
5G line
5 Gauss exclusion zone
risk to implanted devices line
Noise comes from?
Gradient coils
rapid changes in currrent
FDA noise maximum
140 dB (99dB for patients with hearing protection)
Neurostim
induced electrical currents, primarily in extremities
painful
usually high bandwidth readouts and rapid gradient switching
fix = reduce bandwidth, increase TR
SAR
what is?
formula?
basic depends ons
ability to cook a patient via energy of RF pulse
Bo^2 x alpha^2 x duty cycle
Double field strength, quadruple SAR
alpha = flip angle, double = quadruple SAR
Duty cycle inversely related to TR, so double TR = half SAR
Higher SAR sequences?
Spin echo (higher flip angles)
SAR limits
no temp increase greater than 1C
FDA LIMIT 4W/kg
MRI zones
I - outside building
II - No restriction - waiting room - screening takes place here
III - Restricted room - control room, where the techs work.
SHOULD BE A LOCK BETWEEN 2 and 3
IV - Scanner
Code in zone 4?
Techs do CPR in zone 4, move to zone 2 for code team
black hole over left hemiliver
worse with
fix
Dielectric effect ‘local eddy currents’
worse with stronger fields
worse with fat ascitic people
fix- use dielectric pads, parallel RF transmission (SENSE)
Magic angle
better with
Better with longer TE (T2 sequences)
55 degrees
BETTER at higher field strengths
Examples of ‘long, short’ TR and TE
Spin Echo:
- Short TR < 700ms - Long TR > 2000 ms
- Short TE < 25ms - Long TE > 60ms
- For GRE:
- Short TR < 50ms - Long TR > 100 ms
- Short TE < 5 ms - Long TE > 10ms
concentric rings on CT
Ring artifact
defective detector
3rd generation (old) scanners
fix, recalibrate, replace
Lateral res does best with?
effect of gain?
a narrow beam
worse with cranked up gain (widens beam)