RDA MRSI localisation over DICOM T1s

[Chris Adamson]

I have a conundrum with localising MRSI voxels to anatomy. I have 3 datasets each with single slice 16x16 RDA MRSI and DICOM T1 and T2s. One dataset also has the MRSI in Siemens IMA format. The MRS sequence fits nicely with the 1H Brain long echo basis set.

I'm trying to work out the correct location/orientation of the grid overlay and the orientation of the MRS voxels. In the top-right overlay there are a few issues:

1. For the RDA MRS files, it seems to get the row and column spacing wrong, (see tarquin_rda_orientation_row_col.png)

2. The Siemens IMA overlaid onto the localiser has the same problem as the RDA, it seems as if it interprets the row and column spacing incorrectly. (tarquin_ima_overlay.png)

Secondly, as far as the data orientation is concerned. I have attached a TNAA peak magnitude map. Which orientation looks more believable? (tarquin_rda_TNAA.png) The one on the left has high intensity in the CSF, the one on the right in the basal ganglia. The TNAA fit of the DICOM IMA also looks high on the basal ganglia (tarquin_ima_TNAA.png).

Thanks in advance,

Chris Adamson

[Martin Wilson]

Hi Chris,

Thanks for reporting, I think Bhavana has reported a similar issue so
I'm sure between us we can fix this.

I'm a bit confused by some bits of your e-mail so just to clarify:

1) In tarquin_rda_orientation_row_col.png is the left plot the rda and
right plot the ima? Does the ima orientation look correct according to
how it was planned on the scanner?
2) In tarquin_ima_overlay.png - is this the same dataset as shown in
tarquin_rda_orientation_row_col.png except you are using the scout
images rather than the full T1 and T2?

Thanks

Martin

> --
> You received this message because you are subscribed to the Google Groups
> "TARQUIN users' group" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to tarquin_users_g...@googlegroups.com.
> To post to this group, send email to tarquin_u...@googlegroups.com.
> Visit this group at http://groups.google.com/group/tarquin_users_group.
> For more options, visit https://groups.google.com/d/optout.

[Chris Adamson]

1) The left plot is with tarquin's default interpretation of row and column spacing. The right plot is with row and column spacing swapped, so swapping vec[0] and vec[1] in src/common/fidio/CFIDReaderRDA.cpp at line 218. I haven't been able to get access to the scanner console, which is annoying.

2) No the T1 is always used as the background. In this image, it is using the same interpretation of row and column spacing except the top-left coordinate is different. I assume it is something to do with placing the upper left corner at the "center" of the voxel or the corner?

Chris.

[Martin Wilson]

Hi Chris,

I'm fairly sure the ima reader works ok for cubic voxels so it's
possible that the pixel row and col dimensions are being read the
wrong way round for ima.

This could also be true for rda + some other problem (row/col
direction vectors getting confused?), but without knowing how it was
planned it's very difficult to be sure.

Martin

[Chris Adamson]

I will have some time at the scanner with our tech on Monday. I should be able to get more of an idea about the planning then.

Cheers

[Martin Wilson]

Thanks Chris, I'm hoping to make a new release soon so it would be
good to get this sorted next week if possible...

Martin

[Chris Adamson]

I saw our tech this morning and confirmed the location of the MRS voxels. My swapping of the row and column spacing was right. 

Here is a patch file for the source code that will fix Siemens RDA and Siemens IMA. The TNAA fit for the RDA suggests that we need to set the "swap rows and columns" option but that is fine. Martin, can I send you the files so that you have the opportunity to fix the code with some non-cubic voxel data?

[Martin Wilson]

Hi Chris,

Excellent, thanks for tracking that down. Yes, please send example
data if you can as it helps a great deal with debugging.

Martin

[Martin Wilson]

Hi Bhavana,

Greg and I build a Windows release on Wednesday that includes the
patch from Chris (among other things):

https://sourceforge.net/projects/tarquin/files/TARQUIN_4.3.8/

Hope it works for you :)

Martin

On 22 October 2015 at 17:40, Solanky, Bhavana <b.so...@ucl.ac.uk> wrote:
> Hi Martin,
>
> I just wondered how this is going. WOuld be great to analyse some data as soon as possible....?
>
> Also i have some more phantoms acquisitions if that will help this time with a rectangular FOV.
>
> Chris- should i go ahead with the data analysis with your patch?
>
> Thanks
>
> Bhavana
>
> ________________________________________
> From: Martin Wilson <wils...@bham.ac.uk>
> Sent: 12 October 2015 09:38
> To: tarquin_u...@googlegroups.com
> Cc: Solanky, Bhavana
> Subject: Re: [tarquin_users_group] RDA MRSI localisation over DICOM T1s