Centric phase encoding order

[Sarah Needleman]

Hi Tony,

 

I am trying to simulate a sequence with a centric phase encoding order, but when I reorder the rows of the 2D k-space array the reconstructed image looks very different to an image from the same sequence but with sequential ordering.

 

I was wondering if you had simulated a centric sequence and might be able to provide some insight as to what I am doing wrong.

 

The formula I am using to calculate the PE gradient area is ((0.5*FlatTopArea)/(N/2))*((-1)^C)*floor((C+1)/2), and the sample is a rectangle with T1=1300ms, T2=60ms, T2*=10ms. The ‘short TR’ sequence has TR=2.1ms, the ‘long TR’ sequence has TR=5000ms for full relaxation before the next RF pulse.

 

Thank you,

Sarah

[Tony Stoecker]

Hi Sarah,

the centric reordering looks good in the GUI (I looked at the k-space trajectory). However, there are many pitfalls with the simulation of short-TR GRE sequences. I have a couple of suggestions:

1) the short-TR sequence has strong artifacts from spurious echoes. TR>>T2 is sufficient to get rid of them. Otherwise, the sequence needs gradient and RF spoiling: 

Gradient spoiling in read direction (after readout):

<TRAPGRADPULSE Area="1.5*A" Axis="GX" Name="P6" Observe="A=P5.Area"/>

RF Spoiling - Excitation phase:

<HARDRFPULSE Axis="RF" Duration="0.002" FlipAngle="6" Name="P2" InitialPhase="C*(C+1)*50" Observe="C=C2.Counter"/>

RF Spoiling - ADC phase locking:

 <TRAPGRADPULSE ADCs="ADC" Axis="GX" FlatTopArea="2*Pi*1/Dx" FlatTopTime="TE*0.5" Name="P5" PhaseLock="1"

                           Observe="TE=P.TE, ADC=P.Nx, SlewRate=P.GradSlewRate, FOVx=P.FOVx, Dx=P.Dx, PKmax=P.KMAXx"/>

(See also gre.xml sequence in the jemris examples folder. )

2) the "black dots" in the images result from simulating to few spins with T2* in the order of TE. (some spins have strong off-resonance -> no signal)

Try to make T2* >> TE and see if they vanish. Other option is to simulate much more spins such that the effect cancels out per voxel (time consuming) 

3) Image blurring in the long-TR simulation depends on the PSF of the centric reordering. (For linear ordering the steady state is nearly reached in the center of k-space.)

For the short-TR sequence, a couple of preparation pulses without phase-encoding may help. (take care with correctly counting the RF spoiling number)

If TR>>T1, blurring should completely vanish (apparently, TR/T1~3.8 is not completely sufficient). Then, there should be no difference between linear and centric PE.

In summary, simulating with T2>TE and TR>>T2 typically generates nice image. This is not very interesting but a good check, if gradient encoding is correct. Then, with short TR and TE several other effects need to be taken into account: spoiling of spurious echoes / intravoxel dephasing, PSF due to encoding transient signal evolution, ...

Hope this helps,

Tony