Relaxation time correction

[fbri...@gmail.com]

Hi TARQUIN users',

I would like to know if TARQUIN allow to applied correction factors to the metabolite concentrations to account for metabolite T1 and T2 relaxation effects. Otherwise, how to do that?

Thanks in advance,

 

Frédéric

[Mark Mikkelsen]

Hi Frédéric,

If you're scaling to water, this is how TARQUIN applies the T2 correction: http://tarquin.sourceforge.net/user_guide/tarquin_user_guide.html#x1-240007

A fuller description of how to do the relaxation corrections can be found in Gasparovic et al. (2006): http://www.ncbi.nlm.nih.gov/pubmed/16688703


Best,
Mark

[Paul Mullins]

Just to add to what Mark has said,

From my understanding (and Martin can correct me if I'm wrong), TARQUIN by default applies a correction for T2 relaxation of the water signal only.  This is done through the water attenuation parameter (w_att in command line), which is set to 0.7 by default, the same as in LC-Model.  This is an approximate correction, and assumes a short echo press sequence (around TE=30-40ms).  note that TARQUIN also sets the water concentration (w_conc in command line mode) to 35880 mM by default - which assumes the voxel was taken from white matter (again the same as LC-Model).

If you want to correct for the effect of T2 relaxation on the metabolite signals you need to do this posthoc - you also need to be aware of the corrections that have already been applied (water concentration and attenuation as mentioned above).  I take these out of consideration by setting w_conc to 55550 mM and w_att to 1 before fitting the data and then correct after the fact using the method described in Gasparovic et al (2006) taking into account estimates for CSF and gray and white matter in the voxel, (and estimated water concentrations and relaxation for each) along with T@ values for the metabolites from the literature (although at present I am not using tissue specific T2 parameters for metabolites - does anyone know if T2 varies from gray to white matter?).

I have a spreadsheet I use that does this correction (attached) along with some matlab code for partial volume calculations for Philips data (found here http://biu.bangor.ac.uk/projects.php.en - at the bottom).  Georg Oeltzschner (georg.oe...@UNI-DUESSELDORF.DE) has written some code for Siemens which he may share if you ask hime nicely, and there is an implementation of partial volume correction in the GANNET code (http://gabamrs.blogspot.co.uk/)that I think is multi-vendor (but I'm not sure as I have yet to try it out).

Note the T@ values I use in my spreadsheet are average results culled from the literature - I'm still collecting references for T2 of metabolites, so if anyone else on the list has a favourite publication they use let me know I'll add it.  I have started to collate these into an excel spreadsheet (also attached - it's not been updated for a while, but ti does have references) and am always happy when people send me other references as it saves me having to search for them myself. :)

Hope this helps.

Paul.

-----------------------------------------------------------
-----------------------------------------------------------
Paul Mullins PhD
Director, Senior MRI Physicist
Bangor Imaging Center,
Reader, School of Psychology, 
Bangor University.

Adeilad Brigantia
Penrallt Road
Gwynedd LL57 2AS
United Kingdom

Ph: ++44 (0) 1248 383631
email: p.mu...@bangor.ac.uk
-----------------------------------------------------------
----------------------------------------------------------- 

 

Rhif Elusen Gofrestredig 1141565 - Registered Charity No. 1141565 Gall y neges e-bost hon, ac unrhyw atodiadau a anfonwyd gyda hi, gynnwys deunydd cyfrinachol ac wedi eu bwriadu i'w defnyddio'n unig gan y sawl y cawsant eu cyfeirio ato (atynt). Os ydych wedi derbyn y neges e-bost hon trwy gamgymeriad, rhowch wybod i'r anfonwr ar unwaith a dilewch y neges. Os na fwriadwyd anfon y neges atoch chi, rhaid i chi beidio a defnyddio, cadw neu ddatgelu unrhyw wybodaeth a gynhwysir ynddi. Mae unrhyw farn neu safbwynt yn eiddo i'r sawl a'i hanfonodd yn unig ac nid yw o anghenraid yn cynrychioli barn Prifysgol Bangor. Nid yw Prifysgol Bangor yn gwarantu bod y neges e-bost hon neu unrhyw atodiadau yn rhydd rhag firysau neu 100% yn ddiogel. Oni bai fod hyn wedi ei ddatgan yn uniongyrchol yn nhestun yr e-bost, nid bwriad y neges e-bost hon yw ffurfio contract rhwymol - mae rhestr o lofnodwyr awdurdodedig ar gael o Swyddfa Cyllid Prifysgol Bangor. This email and any attachments may contain confidential material and is solely for the use of the intended recipient(s). If you have received this email in error, please notify the sender immediately and delete this email. If you are not the intended recipient(s), you must not use, retain or disclose any information contained in this email. Any views or opinions are solely those of the sender and do not necessarily represent those of Bangor University. Bangor University does not guarantee that this email or any attachments are free from viruses or 100% secure. Unless expressly stated in the body of the text of the email, this email is not intended to form a binding contract - a list of authorised signatories is available from the Bangor University Finance Office.

[Martin Wilson]

Yes, I agree with this. The default assumptions will get you in the ball-park, but it's always going to better to apply your own correction if possible.

The defaults also assume T1 isn't an issue (ie long TR).

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.

[fbri...@gmail.com]

Thanks everybody for your contribution, especially Paul Mullins (which my thesis is based on their work: very interesting articles about glutamate & schizophrenia) for this spreadsheet.
Best,
Frédéric

[Felix Raschke]

Hi all,

I got a few questions about relaxation correction.

Paul thank you very much for the spreadsheets and also for you blog entry about relaxation correction. I finally got around to implemented the Gasparovic correction in Excel and it gives identical results to your spreadsheet when using identical T1s and T2s which is good.

I got MRS data from 17 subjects. For each subject a MRS voxel was placed either in GM in the ACC or in parietal WM. From that voxel location I got a short TE STEAM, PRESS TE68, PRESS TE80, PRESS TE105 and PRESS TE131. TRs vary slightly. Now I thought this would be a good dataset to test the absolute quantification as we would expect all 5 sequences to give very similar NAA, Cr and Cho measures if the T1s and T2s are correct. When I use the water and metabolite T1s and T2s used in this paper (http://onlinelibrary.wiley.com/doi/10.1002/cne.23634/abstract) I get almost twice the concentration of metabolites at TE131 compared to TE14.
When I use Paul's T1 and T2 values results look better, with a bit of a decrease in metabo concentrations in a WM voxel with increasing TE. I attached a spreadsheet with the results for 3 voxel locations.
The main difference seems to be the water T2 values.

http://www.ncbi.nlm.nih.gov/pubmed/16086319 found T2 in GM and WM of around 100ms and 70ms respectively.
http://www.ncbi.nlm.nih.gov/pubmed/10232510 found T2 in GM and WM of around 110ms and 80ms respectively.

Paul where did you get your T2 values from, they seem lower to what I could find so far.

I would also appreciate any comments and/or suggestions about the corrections!

Thanks everyone

Felix


Paul FYI a few papers reporting T1s and T2s of metabolites separately in GM and WM at 3T:
http://www.ncbi.nlm.nih.gov/pubmed/17534907
http://onlinelibrary.wiley.com/doi/10.1002/nbm.713/abstract
http://onlinelibrary.wiley.com/doi/10.1002/mrm.10640/abstract
http://www.ncbi.nlm.nih.gov/pubmed/10232510

[Paul Mullins]

Hi Felix, banks for the references, I think we may have found these, and are preparing a review.  I will dig up where the water T2 values come from, but you are correct, these will have a substantial effect on concentrations.  

Will see what I can sort out tomorrow.  Just flew back from the states this morning... And boy are my arms tired. :)

Sorry for the bad joke, blame the jet lag.

Sent from my iPad

--

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.

<T1_T2_example.xlsx>

[oeltz...@googlemail.com]

Hi Felix,

which T1 and T2 do you use for water in CSF? I found the results of absolute quantification to heavily rely on them, even if there's only ~5-10% compartmental CSF fraction. Needless to say that literature values vary immensely for CSF. Did you also look at GABA by any chance?

Paul, the relaxation time review is a brillant idea.

Cheers,

Georg 

[Felix Raschke]

Hi Georg,

I haven't really found T1/T2 for CSF. For the moment I just used the values Paul has in his correction spreadsheet (T1=4s, T2=200ms). If you got any values/papers/suggestions let me know!
I looks like that the WM/GM T2s appear to be main cause for the large absolute metabo differences at different TEs but I will have a look at the CSF values as well.

Best wishes

Felix

[Paul Mullins]

Hi all,

so I did the digging like I said I would. it turns out there I could not find the reference for the water relaxation values in the spreadsheet. This spreadsheet had a long history and while there may have been a reference originally, it’s long gone.

So I did some digging, and bit more due diligence and found a few more references for the T1 and T2 of water in Grey and white matter -(still gathering data on CSF, if anyone has references, please send them along).

This is the current version of the spreadsheet updated. i also fixed a small error that had crept in where by the TE was not updating correctly. If you put the TE into cell A81 it should update correctly now. (if you use two different TE’s either use separate sheets, or modify the formulas in cells B62,B63and B64.

The T1’s and T2’s for water have been updated as an average of a few different references. All references are on the second worksheet (T1 and T2 references).

Felix, give these new values a go and let me know how it affects your concentrations at the different echo times.

Apologies for the problems the earlier wrong version may have caused. Note, while I think the T2 of metabolites is largely correct, they may change slightly in the future as a result of a review of the literature I am currently undertaking.

Paul.

[oeltz...@googlemail.com]

Hi all,

for CSF at 3T, a T1 around 4000 ms seems reasonable. If I recall it correctly, Chuck Gasparovic used it constantly for CSF in his 2009 quantitative spectroscopic imaging paper, instead of the T1 values obtained from inversion recovery. I found T1 = 4160 ms in Gussew et al. (MAGMA 2012;25(5):321-33), they cited (Lin C, Bernstein M, Huston J, Fain S (2001) Measurements of T1 relaxation times at 3.0: implications for clinical MRA. In: Proceedings of international society for magnetic resonance in medicine. 11, 21–27 Apr 2001, Glasgow, Scotland, p 1391), I don't know if there is a paper following up on it.

Gussew also mention CSF T2 = 500 ms, taken from (Piechnik SK, Evans J, Bary LH, Wise RG, Jezzard P (2009) Functional changes in CSF volume estimated using measurement of water T2 relaxation. Magn Reson Med 61(3):579–586). And there we go, that's 300 ms difference already in T2.

I'll keep looking out for CSF values, those are the ones that I could find within an instant Zotero browse...

Paul, will you include the known GABA relaxation times in your spreadsheet, too? 

Cheers,

Georg

[Paul Mullins]

Hi George, when we get it all collated, yes it will include GABA T2’s as I become aware of them - however, the T2 for GABA (and other J coupled metabolites) should always be treated with caution.

On the issue of CSF - 

Liberman, G., Louzoun, Y., & Ben Bashat, D. (2013). T 1Mapping using variable flip angle SPGR data with flip angle correction. Journal of Magnetic Resonance Imaging, 40(1), 171–180. http://doi.org/10.1002/jmri.24373

reports T1 for CSF at between 6873 and 4184 (female v’s Male) and also quote Chen et al for a CSF T1 of 4163.  So at least 4000 ms, if not longer.

(Chen L, Bernstein M, Huston J, Fain S.Measurements of T1 relaxation times at 3.0 T: implications for clinical MRA. In: Proceedings of the 9th Annual Meeting of ISMRM, Glasgow, Scotland, 2001. (abstract 1391).) 

I also note that the new T1 of grey matter in the spreadsheet is close to that generally assumed for ASL calculations of CBF of 1.3 s (BASIL user guidelines, FSL). 

 Okay, watch this space, as I update the spreadsheet it will make it’s way onto the forum.

Paul.

-- 
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.

 

Rhif Elusen Gofrestredig 1141565 - Registered Charity No. 1141565

[Felix Raschke]

Hi all,

Paul many thanks for your updated spreadsheet. I just had a quick look but noticed that you have a mix of 1.5T and 3T publications for the water T1s and T2s. E.g. both Vymazal1999 and Gasparovic2006 use 1.5T which explains the significantly shorter T1s.

Georg, thanks for the references!

Looking at all the available literature values and the results from my spreadsheet above I am begining to doubt the benfit of relaxation correction in order to obtain metabolite values that are comparable between different (TE/TR) sequences.
Attached an interesting method I have been eyeballing a long time now has been described here to obtain tissue proportions and water T1s and T2s. Maybe this is the way to go? Can apparently be reduced to a 60-90s scan.

best wishes

Felix

On Thursday, December 4, 2014 at 3:46:03 PM UTC, fbri...@gmail.com wrote:

[Paul Mullins]

Thanks for the catch Felix.  Stupidity on my part.  I've corrected this now.  I think I also need to find a lot more T2 measure data and add that to the spread sheet, as the T2 of water is probably having the most effect when trying to correct for relaxation effects in most cases.  Of course, given the paper and poster you have supplied, it might be important to check how that T2 data was collected.

The attached methods looks interesting.  Am I reading it right in that it if you start at about a TR of 3 secs, and then collect your data with longer TR for each TE you avoid potential T1 effects when collecting T2 data?  Of course you could also get away with using a very long TR to start with, but that has a time penalty.  I'm wondering if it is possible to collect the data without modifying the TR as you modify TE and account for the extra saturation effect if you know the T1?  (Note the RRAMSC method itself does not seem to give T1, a separate saturation recovery experiment would be needed for that, with HSR being used in the paper supplied)

Obtaining the exact tissue water content and T1 and T2 is definitely a better way to go for relaxation correction, but it carries with it a time penalty (doesn't everything) which can be a problem when your MRS data is being collected along with a bunch of other imaging data.  This is one of the supposed benefits of the TE average and 2D type methods of acquisition, - you get relaxation information at the same time as your data, but SNR may be lower for several metabolites, and J modulation can make interpretation of decay rates tricky. (and 2d methods often have their own time penalty).  The Gasparovic paper I cited actually did this, but it took a lot of time to acuire the data.  I think there are now faster methods out there to do those T1 and T2 measures though, so it may now be possible to get a better approximation.  

Note even with this you would still need to do some form of relaxation correction to compare different TE/TR combinations, you would just hopefully have a better correction factor for each individual subject.

Thanks again for the methods papers by the way, very useful.  If anyone else has any papers on T2 and T1 measures etc, please do share, I'm collecting as many as I can get my hands on, but it is mostly happening in my spare time at present, so slow going.  

Paul.

--

[Pom Sailasuta]

Hi Paul,

Here is another T2 measurement for your excel spreadsheet.

Pom

--

Pom

[Paul Mullins]

Thanks Pom, 

We have this one in our larger database, which will be available in the near future.  It shows nicely how you can use TE average data to get T2 measures.

[Andreas Bartsch]

Hi,

this may be of interest:

Ehses, Philipp; Gulani, Vikas; Jakob, Peter M.; Griswold, Mark A.; Breuer,
Felix A. Rapid 3D Relaxation
Time and Proton Density Quantification Using a Modified Radial IR TrueFisp
Sequence. . Proc. Intl. Soc.
Mag. Reson. Med. 18 (2010), #2952

and "Generating Multiple Contrasts Using Single-Shot Radial T1 Sensitive
and Insensitive Steady-State Imaging" MRM 2014 can be extended to generate
T1/T2 relaxation maps quite rapidly.

Cheers,
Andreas

[Paul Mullins]

Thanks Andreas, very useful.

[Felix Raschke]

FYI, I just had a look at the Gasparovic2009 paper that Georg mentioned and although they use 3T data they clearly use water T1s and T2s obtained at 1.5T. They also suggest a CSF T2 of 2.5s which seems pretty long.

Andreas thanks for the papers. Have you used the techniques to create T2 (T1) maps and do you have any experience using those for MRS relaxation correction?

many thanks

Felix


On Thursday, December 4, 2014 at 3:46:03 PM UTC, fbri...@gmail.com wrote:

[Paul Mullins]

Hi Felix, not sure where you get 3T data from the Gasparovic 2009 paper, that work was all done at 1.5 T.  (the 3T was in the process of being installed when the data was collected, and while T2 estimates for water at 3T are mentioned once in the discussion I think this was just to highlight the variability in reported values).  The T2 values for CSF were actually calculated from one subjects data in that study (ventricles where only included in the T2 mapping images for one of the subjects).

Paul.  

--
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.

 

Rhif Elusen Gofrestredig 1141565 - Registered Charity No. 1141565

[Andreas Bartsch]

Hi Felix,

I have used these techniques (I happened to be involved in the second which we are trying to improve) but not (yet) for MRS relaxation correction.

If you are interested, you may email me offlist.

Cheers,

Andreas

--

[Neda Naderi]

Dear All

Hi,

I read the whole conversation but I somehow couldn't reach a conclusion for W att factor.

I have CSI GE data with TE=144 at 3T and when I calculated the factor according to tarquin formula I got the number=0.2940

I selected 80 ms for water T2 and 250 for metabolites..just a rough estimate according to:(https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3852663/)...why did I get this number???...As we expect to have shorter T2 by increasing the magnetic field and more T2 effect at higher TE..both means more signal loss so we need a higher number to compensate...

I would appreciate your reply.

Regards,

Neda

[Paul Mullins]

Hi Neda,

 

The purpose of the water attenuation factor in Tarquin is to account for the fact that the water signal will have attenuated with longer TE. Basically it is an estimate of how much water signal is left at the TE chosen. This means that as TE gets longer, the attenuation factor gets smaller (as you found out).

 

You can use your estimate, or you can also just set the attenuation factor to 1, water concentration to 55500 mM – and correct for relaxation effects outside of Tarquin using excel and the equations discussed in:

“Near, J., Harris, A. D., Juchem, C., Kreis, R., Marjańska, M., Öz, G., Slotboom, J., Wilson, M., & Gasparovic, C. (2020). Preprocessing, analysis and quantification in single-voxel magnetic resonance spectroscopy: Experts’ consensus recommendations. NMR in Biomedicine, n/a(n/a), e4257. https://doi.org/10.1002/nbm.4257”

Or other papers that deal with this issue.

 

Note, if you are wanting to truly account for all relaxation effects, you also need to consider the T2 and T1 of the metabolites of interest.

 

Paul.

 

Prof Paul Mullins Ysgol Gwyddorau Dynol ac Ymddygiadol E-bost: p.mu...@bangor.ac.uk Ffôn: 01248 383631 Rydw i’n siarad rhywfaint o Gymraeg	Prof Paul Mullins Professor in Psychology School of Human and Behavioural Sciences Email: p.mu...@bangor.ac.uk Phone: 01248 383631 I can speak some Welsh

If you are a student wishing to have a tutor or other meeting, please use the link below to book my office hours

https://outlook.office365.com/owa/calendar/ProfessorP...@bangoroffice365.onmicrosoft.com/bookings/

To view this discussion on the web visit https://groups.google.com/d/msgid/tarquin_users_group/742c7e9d-e26b-46a2-8627-54763dbc645cn%40googlegroups.com.

 

Mae croeso i chi gysylltu gyda'r Brifysgol yn Gymraeg neu Saesneg You are welcome to contact the University in Welsh or English

[Neda Naderi]

Thank you very much Paul,  ..what about 55550 and the estimated number with  my parameters??...