Support needed for GE MEGA-PRESS spectroscopy

[Bartosz Kossowski]

Dear Users,

together with my colleague we've been trying to correctly read GE pfiles with MEGA-PRESS single-voxel spectroscopy. Those files were acquired on Discovery 750w ver. 25.1 with an original Ralph Noeske's work-in-progress sequence (standard parameters).

According to the manual we have tried to set up different subtraction option but the attached, preprocessed spectrum does not look reasonably. Please let me know what is the appropriate parameter set for that type of data. I can provide the original pfile upon request.

Bartosz Kossowski

[Ben Rowland]

Hi Bartosz,

That is an amazing coincidence, I was just working on my software tool (OpenMRSLab) for doing preprocessing of GE pfiles (including MEGAPRESS) for Tarquin quantification when I received your message. I have done some testing of my software on the data I have but if you could let me have your pfile I will make sure that it works with my code and will then help you get it working. OpenMRSLab is all open source and freely available from GitHub, but the pfile modifications haven’t been uploaded yet as I only got them working today, and I still need to get some supporting documentation in place.

Ben

Ben Rowland DPhil

Lecturer in MR Imaging Physics

Division of Cancer Sciences

University of Manchester

Email: Benjamin...@manchester.ac.uk

--
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 view this discussion on the web visit https://groups.google.com/d/msgid/tarquin_users_group/a62b0c3e-c2e8-4686-9f57-2ca207626154%40googlegroups.com.
<plotgaba.pdf><plotgabafit.pdf>

[Ben Rowland]

Hi Bartosz,

I had a look at your data which comes up a bit noisy but not too bad. Attached is a screenshot of the data and fit from Tarquin after initial processing through OpenMRSLab. There is a pretty substantial MM peak at 1.7ppm that Tarquin doesn’t seem to take into account, someone more familiar with Tarquin’s MEGAPRESS handling will have to comment on that. How does this compare with your GANNET processing? The attached concentrations are with wref scaling but as the Tarquin manual says this is probably not right for MEGAPRESS, although it probably makes it more consistent across scans at least.

Let me know if you want me to help you get this processing pipeline set up for your site.

Ben

To view this discussion on the web visit https://groups.google.com/d/msgid/tarquin_users_group/CC0BB39A-47BD-45AB-AEF7-E85158A0D164%40mac.com.

[Bartosz Kossowski]

Ben,

the fit looks very promising. Absolute concentrations does not correspond to Gannet but as far as I understand they just need to be rescaled linearly. 

I would appreciate Your help to combine OpenMRSLab & Tarquin. Could You please send me some tutorial or the script You’ve already tested? I have much more data from GE and Siemens (TWIX) as well and I would love to compare those results in another software. 

B. 

You received this message because you are subscribed to a topic in the Google Groups "TARQUIN users' group" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/tarquin_users_group/NtJc6I7TWNY/unsubscribe.
To unsubscribe from this group and all its topics, send an email to tarquin_users_g...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/tarquin_users_group/F217D4F7-8F2D-4A4E-9C14-FC0EF35F8837%40mac.com.

[Helge Zöllner]

Hi Bartosz,

you could also use the new MATLAB MRS toolbox Osprey (https://www.biorxiv.org/content/10.1101/2020.02.12.944207v1.full#fn-5). It features pre-processing of all major vendors and various sequences, as well as its own linear-combination modeling algorithm. The processed data can be exported as vendor native files (In case of GE you would need to use the jMRUI text option, which you can load in Tarquin afterwards) to be quantified in other tools (Tarquin,LCModel or jMRUI). It allows batch processing and there are examples on how to set up a processing pipeline on GitHub (https://github.com/schorschinho/osprey).

Regarding the 1.7 MM peak, it is just not considered in Tarquin's current MEGA-PRESS model. It only works with the 8 peaks mentioned in Ben's output.

I hope this helps.

Cheers,

Helge

Am Donnerstag, 7. Mai 2020 12:08:07 UTC-4 schrieb Bartosz Kossowski:

Ben,

the fit looks very promising. Absolute concentrations does not correspond to Gannet but as far as I understand they just need to be rescaled linearly. 

I would appreciate Your help to combine OpenMRSLab & Tarquin. Could You please send me some tutorial or the script You’ve already tested? I have much more data from GE and Siemens (TWIX) as well and I would love to compare those results in another software. 

B. 

W dniu czw., 7.05.2020 o 16:42 'Ben Rowland' via TARQUIN users' group <tarquin_u...@googlegroups.com> napisał(a):

Hi Bartosz,

I had a look at your data which comes up a bit noisy but not too bad. Attached is a screenshot of the data and fit from Tarquin after initial processing through OpenMRSLab. There is a pretty substantial MM peak at 1.7ppm that Tarquin doesn’t seem to take into account, someone more familiar with Tarquin’s MEGAPRESS handling will have to comment on that. How does this compare with your GANNET processing? The attached concentrations are with wref scaling but as the Tarquin manual says this is probably not right for MEGAPRESS, although it probably makes it more consistent across scans at least.

Let me know if you want me to help you get this processing pipeline set up for your site.

Ben

On 7 May 2020, at 13:44, 'Ben Rowland' via TARQUIN users' group <tarquin_u...@googlegroups.com> wrote:

Hi Bartosz,

That is an amazing coincidence, I was just working on my software tool (OpenMRSLab) for doing preprocessing of GE pfiles (including MEGAPRESS) for Tarquin quantification when I received your message. I have done some testing of my software on the data I have but if you could let me have your pfile I will make sure that it works with my code and will then help you get it working. OpenMRSLab is all open source and freely available from GitHub, but the pfile modifications haven’t been uploaded yet as I only got them working today, and I still need to get some supporting documentation in place.

Ben

Ben Rowland DPhil

Lecturer in MR Imaging Physics

Division of Cancer Sciences

University of Manchester

Email: Benjami...@manchester.ac.uk

On 7 May 2020, at 13:27, Bartosz Kossowski <bartosz....@gmail.com> wrote:

Dear Users,

together with my colleague we've been trying to correctly read GE pfiles with MEGA-PRESS single-voxel spectroscopy. Those files were acquired on Discovery 750w ver. 25.1 with an original Ralph Noeske's work-in-progress sequence (standard parameters).

According to the manual we have tried to set up different subtraction option but the attached, preprocessed spectrum does not look reasonably. Please let me know what is the appropriate parameter set for that type of data. I can provide the original pfile upon request.

Bartosz Kossowski

--
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_group+unsub...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/tarquin_users_group/a62b0c3e-c2e8-4686-9f57-2ca207626154%40googlegroups.com.
<plotgaba.pdf><plotgabafit.pdf>

--
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_group+unsub...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/tarquin_users_group/CC0BB39A-47BD-45AB-AEF7-E85158A0D164%40mac.com.

--
You received this message because you are subscribed to a topic in the Google Groups "TARQUIN users' group" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/tarquin_users_group/NtJc6I7TWNY/unsubscribe.

To unsubscribe from this group and all its topics, send an email to tarquin_users_group+unsub...@googlegroups.com.

[Martin Wilson]

Hi Bartosz,

I think the latest version of the code has a fix for the most recent
GE P format, but this hasn't been released as a binary - so you'd need
to recompile (this is easiest on Linux). Let me know if you'd like to
try, otherwise using some other software to convert to a simpler data
format (as suggested by others) is a good idea.

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.

[Bartosz Kossowski]

Hi Martin,

so I compiled the master version from https://github.com/martin3141/tarquin. However, the results are the same as with the official release. Despite changes od averaging options between "all, even and subtract", none of them resembles Ben's result.

B.

W dniu niedziela, 10 maja 2020 10:18:06 UTC+2 użytkownik Martin Wilson napisał:

Hi Bartosz,

I think the latest version of the code has a fix for the most recent
GE P format, but this hasn't been released as a binary - so you'd need
to recompile (this is easiest on Linux). Let me know if you'd like to
try, otherwise using some other software to convert to a simpler data
format (as suggested by others) is a good idea.

Martin

On Thu, 7 May 2020 at 13:27, Bartosz Kossowski
<bartosz....@gmail.com> wrote:
>
> Dear Users,
> together with my colleague we've been trying to correctly read GE pfiles with MEGA-PRESS single-voxel spectroscopy. Those files were acquired on Discovery 750w ver. 25.1 with an original Ralph Noeske's work-in-progress sequence (standard parameters).
> According to the manual we have tried to set up different subtraction option but the attached, preprocessed spectrum does not look reasonably. Please let me know what is the appropriate parameter set for that type of data. I can provide the original pfile upon request.
> Bartosz Kossowski
>
> --
> 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_group+unsub...@googlegroups.com.

[Martin Wilson]

Hi Bartosz,

Thanks for checking, I can take a closer look if you’re able to send me the p file.

Martin 

On Tue, 12 May 2020 at 14:57, Bartosz Kossowski <bartosz....@gmail.com> wrote:

Hi Martin,

so I compiled the master version from https://github.com/martin3141/tarquin. However, the results are the same as with the official release. Despite changes od averaging options between "all, even and subtract", none of them resembles Ben's result.

B.

W dniu niedziela, 10 maja 2020 10:18:06 UTC+2 użytkownik Martin Wilson napisał:

Hi Bartosz,

I think the latest version of the code has a fix for the most recent
GE P format, but this hasn't been released as a binary - so you'd need
to recompile (this is easiest on Linux). Let me know if you'd like to
try, otherwise using some other software to convert to a simpler data
format (as suggested by others) is a good idea.

Martin

On Thu, 7 May 2020 at 13:27, Bartosz Kossowski
<bartosz....@gmail.com> wrote:
>
> Dear Users,
> together with my colleague we've been trying to correctly read GE pfiles with MEGA-PRESS single-voxel spectroscopy. Those files were acquired on Discovery 750w ver. 25.1 with an original Ralph Noeske's work-in-progress sequence (standard parameters).
> According to the manual we have tried to set up different subtraction option but the attached, preprocessed spectrum does not look reasonably. Please let me know what is the appropriate parameter set for that type of data. I can provide the original pfile upon request.
> Bartosz Kossowski
>
> --
> 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 view this discussion on the web visit https://groups.google.com/d/msgid/tarquin_users_group/a62b0c3e-c2e8-4686-9f57-2ca207626154%40googlegroups.com.

--

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 view this discussion on the web visit https://groups.google.com/d/msgid/tarquin_users_group/cbf1cfd6-4990-48f4-954c-85f92c6df1d9%40googlegroups.com.

[Martin Wilson]

Hi Bartosz,

You need to set the --inv_even_pairs option to true as the phase
between the water reference scans is 180 degrees different for some
versions of Ralphs sequence, the following should get you started:

tarquin --input 5_1015_GABA_te68_tr1800.7 --output_pdf plot.pdf
--int_basis megapress_gaba --inv_even_pairs true --dyn_av subtract

Martin

[Surya Gnyawali]

Hi all,

I liked the idea of osprey.m. I have Bruker seq format MRS data. But Osprey doesn't read that format. Any idea what file format is possible and how to convert.

Thanks

Regards

Surya

[Surya Gnyawali]

Hi Bartosz,

Is bruker's PRESS MRS data not readable in Osprey? 

I tried, either I could not follow or it doesn't. Please Advise.

Thanks

Surya

[Helge Zöllner]

Hi Surya,
Hi Bartosz,

as described in the readme.md Osprey is currently not supporting Bruker data. However, it would be possible to implement this into Osprey. In the interest of not hijacking this thread or forum for Osprey related questions I will point you to the https://forum.mrshub.org/c/mrs-software/osprey/10 forum. You could consider posting your request there and we can discuss how to move this forward.

Thanks

Helge

[Bartosz Kossowski]

Dear All,

I wrote an interface for Gannet in version 2 to read Bruker data some time ago. It's a bit nasty but You may take it as a starting point - https://github.com/bkossows/gannet-bruker.

B.

[Helge Zöllner]

Hi Bartosz,

That's a great suggestion. As Osprey's data format is based on the FID-A structure I'd try to develop the functions based on the input/output functions for Bruker which are already implemented in FID-A.

H.