ICE recon for DTI

[mkulam]

Hi,

I wrote a diffusion weighted EPI sequence with partial Fourier which reconstructs the image on the scanner using ICE STD when I acquire a single volume. However, when I have more than one direction in the same scan the scanner stops after the first volume. Any idea what maybe causing this?

Thanks,
Magdoom

[m-a-x-i-m-z]

Hi,

which data labels / counters do you use for these additional directions?

Best regards,
Maxim

[mkulam]

I use SET for additional directions.

[m-a-x-i-m-z]

Then you should explore the LogViewer output and try to understand why the recon crashes.
--M.

[dabosch]

Funny, I happened to stumble over a very similar issue with a (completely) different sequence today. I'm trying to use openrecon with a custom sequence. ICE is very picky about labels being correct and not violating any limits communicated to it earlier.

@m-a-x-i-m-z , can you have a look at the Siemens interpreter (v1.5.2) please?
I think in line 2683 of Arbitrary.cpp there's an error:

        if (m_b2D_Mode)
        {
            if (u_bSwapSlcAndSet)
                return MRI_SEQ_SEQU_ERROR; // no swap of counters in the 2D mode            
        }
        else
        {
        //...
		if (u_selDataExportMode == recon2D || u_selDataExportMode == recon3D || u_selDataExportMode == reconSTD || u_selDataExportMode == gadgetron2D)
            pSeqExpo->setNSet(arbitrarySBB.getLabelStorage().maxVal(SLC) + 1);
        //...
       }

the way I read this code, if the 2D mode is activated, the number of sets is not communicated to the reconstruction.
A simple fix would be to add two lines and modify it to:

        if (m_b2D_Mode)
        {
            if (u_bSwapSlcAndSet)
                return MRI_SEQ_SEQU_ERROR; // no swap of counters in the 2D mode
            if (u_selDataExportMode == recon2D || u_selDataExportMode == recon3D || u_selDataExportMode == reconSTD || u_selDataExportMode == gadgetron2D)
                pSeqExpo->setNSet(arbitrarySBB.getLabelStorage().maxVal(SET) + 1);
        }
        else
        {
/...

Before I wasn't able to run 2D sequences with multiple sets. With this change the sequences run and ICE reconstructs them.

Best
Dario

[m-a-x-i-m-z]

Hi Dario,

yes, indeed, we don't set SET dimension size in the 2D mode, good spot!

--M.

[mkulam]

Thanks Dario for the catch. FYI, LogViewer reports failed due to invalid loop counters. There seems to be an error in IceProgram. Another error is reported in block 19274 of ArbitrarySBB

[m-a-x-i-m-z]

ICE error should contain ICE object boundaries -- these are the maximum permissible counters as announced by the sequence. If cset in the boundary struct is equal to 1, then this is the case.

[mkulam]

cset is indeed set to 1

[m-a-x-i-m-z]

I've just created an issue to be able to track the problem better. See it here
--M.