Background
Fat spins slower than water by about 3.4 ppm
water is about 4.7 ppm
fat is about 1.3ppm
http://www.auntminnie.com/index.aspx?sec=sup&sub=mri&pag=dis&ItemID=68144
whats up with ppm?
see discussion
I found the following summary in my files. (May have been from Jeff Weinreb currently at Yale) Thanks to John Choi '18 for transcribing.
What are the three types of fat saturation?
Describe briefly how they work.
1) Frequency-selective saturation: the RF pulse is centered on the fat resonance, nulling the Z component of fat magnetization. An irreversible process dependent on T1 relaxation rates.
Frequency-selective excitation: a pair of RF pulses are designed to leave the fat magnetization longitudinal while maintaining a transverse X-Y component for water. Unlike #1, no spatially selective slice or slab selection is employed.
2) The second type of fat saturation uses an inversion recovery sequence. An inversion recovery sequence is a 180 degree pulse followed at a time TI by a spin-echo (90-180) pulse.
The TI is set at the null point of fat (the point where the T1 curve of fat crosses the X axis). At this point, fat has no longitudinal magnetization to be detected by the spin-echo pulse.
FLAIR is a related technique where the TI is set to the null point of water. Therefore, CSF signal is suppressed.
3) The third technique is a Dixon-type technique where the fat and water protons in the same voxel are out of phase and cancel each other out. This occurs every 2.2-2.3 milliseconds on our 1.5 T machines. Therefore, if the TE is 2.2, 4.4-4.6, 6.6-7.2, etc. the fat should be cancelled. This effect is reduced as the TE is lengthened since the fat protons are not perfectly homogeneous (In order to get a TE this short, we typically use a gradient echo technique.)
so at 1.5T that is
(ppm)(gyromagnetic ratio)(field strength)
(3.4/1000000)42.576MHz/T)(1.5T)=217 Hz this is why you may see the statement that fat and water are different by 220 Hz
or at 3T
about 435 Hz
In phase/out of phase (India ink) aka chemical shift of the second kind https://mriquestions.com/chemical-shift-2nd-kind.html
When the 90 degree pulse is given, fat and water are in phase, since the frequency of fat is slower than water, eventually, the vectors are 180 degrees out of phase and cancel each other out. Later the water catches up to the fat (like a runner getting lapped in a track event, or the minute hand overlapping the hour hand yet again) and they are in phase again.
Why are fat and water out of phase at TE 2.3 ms (at 1.5T) ?
This is related to the question.
What is the interval between the hour and minute hands aligning on a clock?
You take the reciprical of the difference in frequencies
1/(1/60-1/(12x60))= or 65m 27 sec
alternate way of figuring this out
http://wiki.answers.com/Q/How_many_times_do_a_clock's_hands_overlap_in_a_day
Lets use 220 Hz as the difference between fat and water.
1/220 Hz=4.55ms is the time interval for when fat and water are in phase. So at half that time or about 2.3 ms fat and water are out of phase
0 in phase
2.3 out of phase
4.6 in phase
6.9 out of phase
note this effect is less apparent for longer TEs since the population of spins dephases. (The clock analogy no longer holds for long TEs.)
what would happen at 3T
1/435 Hz is 2.3ms
0 in phase
1.15 out of phase
2.3 in phase
3.45 out of phase
4.6 in phase
5.75 out of phase
6.9 in phase
etc. etc.
Note At 1.5T we use this to try to suppress fat by setting the TE at 6.9.
Would this be a good strategy at 3T?
No, fat and water are in phase at 6.9 seconds
Chemical shift
The MRI image is created assuming that all protons are water with the same frequency. From the discussion above, we know that fat spins slower than water. This causes misregistration of fat with respect to water in the frequency encoding direction. The amount of shift can be determined as the number of pixels of shift as noted below:
The receiver bandwidth is the frequency range that is used over the field of view (FOV) for the image in question.Bandwidth (BW) is typically expressed as +/- kiloHertz. So the range of frequency for the image in Hz is:
2(BW)(1000Hz/kHz)
At 1.5 T fat shift is about 220 Hz.
The amount of shift (in pixels) can be determined by:
pixels of fat shift= 220(matrix)/((2)(BW)(1000Hz/kHz))
pixels of fat shift=0.11(matrix)/(BW)
BW [kHz]
What would be the equation for 3 T?
pixels of fat shift=0.22(matrix)/(BW)
This is the derivation of the equation in the reference.
One way to reduce chemical shift artifact is to increase the bandwidth. This can reduce the signal to noise. This strategy is used at 3T since there is more signal to noise at 3T.
One must be care when trying to increase signal to noise by reducing bandwidth, since unacceptable chemical shift artifacts will result.
Q: Which way does fat shift?
A: Fat shifts in the direction of decreasing frequency encoding gradient. For our GE scanners, Increasing frequency gradient is given by the right hand rule. If you stand at the foot end opening of the magnet, point your middle finger toward the other end, your index finger will point up and your thumb points to the patients right. So depending on what the phase frequency orientaton is, fat should shift inferiorly or to the left in the axial plane, inferiorly or posteriorly in the sagittal plane and inferiorly or to the left in the coronal plane.
For routine imaging, the slice is also encoded with a gradient which can cause misregistration of fat out of the slice as well. See the 1991 article by Smith
http://www.ncbi.nlm.nih.gov/pubmed/1887036
Reference:
This shows examples of fat shift using various bandwidths.
Fat saturation (Chemical saturation/Spectral saturation)
Since fat spins slower, a presaturation pulse covering the frequency of the fat peak is given before the
regular imaging sequence. This works better at higher field than at lower field be cause there is more separation (in Hz as ppm is always the same) in the peaks.
e.g. 0.3T 44 Hz
1.5T 220 Hz
3.0 T 440 Hz
What is a presaturation pulse?
This is a technique to suppress signal from selected tissues. It is a 90 degree pulse given before the imaging sequence. As the imaging sequence progresses, the presaturated tissues are 90 offset from the rest of the tissues. When the wanted protons are in the transverse plane giving dignal, the unwanted protons are in a "vertical" orientaion and contribute no signal to the image.
There are 3 common uses.
Spatial presaturation
In plane: part of the image is presaturated. for example on spine images, we presat the abdomen so that artifact does not spread into the spine
Out of plane: for example in TOF MRA of the neck, the slice superior to the imaged slice is presaturated to eliminate the venous signal
Fat Saturation
The whole image is presaturated with a pulse that only presats out the fat.