Can we import fields/coils in Koma?

[Anais-A]

Hello,

Thank you for developing such an exciting simulation tool.

I am considering using Koma for my project, but I would need to be able to match the coil I am using on the real scanner. For that, I can provide field maps (B1+ B1-, transmitting and receiving E fields) matching my coil. However, I could not find any functionality in Koma to change the coil parameters.

Would it be possible for me to simulate my coil in Koma?

Thanks,
Anais

[cncastillo]

Hi Anais!

Thanks for taking the time to use our simulator 😄!

Currently, we have yet to support it (see #150 ). The main reason is that I haven't needed it, so we have prioritized other aspects, like simulation speed and Pulseq-compatibility.

Having said that, it is totally possible to implement it, and it should be relatively easy, as I have been preparing the code to include coils. I assume you are obtaining your B1 maps from electromagnetic simulations. If so, please tell us more about the formats in which you get the B1 fields. It could be helpful for other people to import them directly from their simulation software.

We would need to incorporate the coils in our sys::Scanner type. Then, the required modifications are:

1. To add coils to the Scanner type:
   https://github.com/cncastillo/KomaMRI.jl/blob/6469a104c595917237d2c2cca31999d25590ea70/KomaMRIBase/src/datatypes/Scanner.jl#L31-L46

2. Change the simulation output dimensions (the 1 should be replaced by the number of receive coils):
   https://github.com/cncastillo/KomaMRI.jl/blob/6469a104c595917237d2c2cca31999d25590ea70/KomaMRICore/src/simulation/Bloch/BlochSimulationMethod.jl#L11

3. To multiply Mxy by the coil sensitivities to obtain a different signal per coil:
   https://github.com/cncastillo/KomaMRI.jl/blob/6469a104c595917237d2c2cca31999d25590ea70/KomaMRICore/src/simulation/Bloch/BlochSimulationMethod.jl#L60

4. Apply a different B1 per coil:
   https://github.com/cncastillo/KomaMRI.jl/blob/6469a104c595917237d2c2cca31999d25590ea70/KomaMRICore/src/simulation/Bloch/BlochSimulationMethod.jl#L90

I believe that in this case, what we call B1 is actually $B_{1}^{+}$, and $B_{1}^{-}$ is the one responsible for inducing an $E$-field and therefore only relevant for SAR calculations but not for the simulation. Is this correct?

It would be really nice if you helped us to understand your needs to implement this properly 😁.

Cheers!

[Anais-A]

Hi Carlos,

That is good news.
Sorry for the files, you should be able to access them now. Can you please confirm that you can read them?
I will discuss what you propose with my team but it sounds good to me.

[cncastillo]

Yes I can read them now :) Let me know what you think after talking with your team. Cheers!

[Anais-A]

Hi Carlos,

I met with my team and we agreed that it would be helpful if you could provide means to import custom coil sensitivities in the compressed format I gave you or in a format we can derive from the mat file (or the mat file itself).

To do so, we propose that you simulate a uniform sphere and see the results that should be easily interpreted for our birdcage coil sensitivities (we are not considering motion) that I already gave you. We can review the results if you need a double-check with what we have on our side.

For the transmit part, we do not need it for now, but we'll be happy to test it when you implement it.

Also, reading your answers and your code, we had a few questions, to make sure we understand what you are doing:

• Currently, are you using B1+ for transmit? it seems you are not taking any sensitivities into account.
• Do you assume that the flip angle is uniform in every voxel?

Thank you,
Anais

[cncastillo]

It is a great idea to simulate a uniform sphere. Could you point me to the grid resolution of the B1 maps in the files you shared?

Currently, we are implicitly assuming constant coil sensitivities and a spatially uniform flip angle. That is why you do not see coil sensitivities and B1 maps in the code, but that will change with the modifications we want to do.

Hope that clear some things up,
Cheers!

[Anais-A]

Hi Carlos,

Thank you for the explanations, it is much clearer now.

The grid resolution of the files I already gave you is 5x5x5mm^3, voxel isotropic. However, these were tailored for a head model. Here are new maps at 3T, with a resolution of 1x1x1mm^3, adapted for the sphere example: https://drive.google.com/drive/folders/1fPxMM5eLSFGLwEhmY1ALLJa9Ca_T5MMp?usp=sharing
Our sphere uses these parameters:

• Radius: 10cm
• Relative permittivity: 70
• Conductivity: 0.5 S/m

Best,
Anais

[Anais-A]

Hi @cncastillo ,

Any progress on this end?
We are eager to try!

[cncastillo]

Hi @Anais-A !

Sorry, we were a little busy working on:

• KomaMRIFiles: RF phase read with the opposite sign from Pulseq #315
• Hot fix empty grad #316
• [BUG] CodeCov Not Working #335
• Incorrect gradient interpretation when waveforms do not end in zero #321.

When #321 is finished, @beorostica will start working on the multi-coil support 😄.

[Anais-A]

Thank you for the news :)

Please tell me if I can help when you will be handling it.

[cncastillo]

https://juliahealth.org/KomaMRI.jl/previews/PR454/tutorial/06-Coils/