TR- Reason says long practice says short

Hi,

I’ve been reading and thinking about TR recently, from what I gather longer TR (~ 4 or 5s) is recommended to eliminate T1 relaxation effects. This seems to be recommended for quantitative MRS and provides a high steady-state Mz. I understand that this reduces time efficiency in studies, but so much of the MRS literature is done with a TR of 2s.
Would the report of ratios or mM affect how one interprets this?
I am interested to hear what others think when they see MRS data collected at a short TR and some suggestions for future study designs!
Thanks!

Hi Jess,

The 2s for 1H-MRS is pretty much a canonized trade-off. Yes, in an ideal world, you want to measure at as long a TR as possible, so that the different metabolite signals don’t get weighted by their individual T1s.
The enemy you’re up against in the in-vivo is SNR, since you’re constrained by the maximum duration of the experiment. SNR grows with the square root of the number of averages, as you know. It also grows with TR (as more longitudinal magnetization recovers between shots), but with diminishing returns for TR >> T1.

So you’re looking for a ‘sweet spot’ TR affording you enough ‘SNR per unit time’. The optimal TR to optimize that metric is super-short with a low flip-angle, but at the expense of introducing a hell of a lot of T1 weighting. TR = 2s is a trade-off between all of these considerations. One might argue that the T1 weighting resulting from that is still large enough to make it difficult to compare metabolite estimates (and they’d be right, it’s one of the main reasons no one really likes to call their estimates “concentrations”). But as long as we don’t have a surefire way to measure individual T1 reliably and fast enough, we have to pick a can of worms to open. It’s the reason why I’m personally really excited about multi-parametric MRS (essentially a fingerprinting technique where instead of repeating the same sequence 64, 128, or 300 times, you vary TE, TR and the flip angle between each average, based on a pre-optimized schedule, and then model the results to extract estimates for T1/T2). It might not solve everything, but at least make the worm dish a bit more palatable (read: if we have to do relaxation correction, it’s probably better to use individual estimates than literature values).

This might be a bit lengthy and you might have guessed it already, but maybe this was helpful.
Best,
Georg

2 Likes

The field strength would also matter in this case. T1 relaxation times increase at higher field strength (7T or above), so T1 weighting would be a bigger factor at UHF over 3T or 1.5T. Luckily, we see SNR gains in the higher field strength, so fewer transients would be needed.

1 Like

Hi All,

I hesitate to support the idea that a TR of 2s is canonized in the MRS field. I think it is common, but not a standard. The longer TRs of 4s or longer at 3T offer you the option to perform calculations to correct for a smaller amount of T1 weighting (combined with correcting for T2 and water content) to obtain absolute concentrations in many cases. As @admin mentions, a major issue is whether there exists good literature values of the metabolite relaxation times and water content for the field strength, region of interest, and study population from which you have obtained MRS data to use as correction factors in the calculations. It would be great to not have to use literature values, but if you have a TR of 4s or longer at 3T, it reduces the influence of the T1 weighting correction factor on your absolute concentration values so the literature values have less of an impact on your resulting concentration values. Whereas if your TR is only 2s, then the T1 weighting correction factor had a greater influence so you really do need to have individually measured estimates as opposed to literature values.

Note about multi-parametric MRS: @admin I’m sure you’re aware of the paper Two-dimensional linear-combination model fitting of magnetic resonance spectra to define the macromolecule baseline using FiTAID, a Fitting Tool for Arrays of Interrelated Datasets . FiTAID is a tool developed by Roland Kreis’ group to analyze various types of multi-parametric MRS datasets. It might be a good idea to reach out to Roland to see what is the latest status of FiTAID.

Note about 7T: I agree with @ywpark about gaining back some SNR at 7T. 7T and higher also usually requires longer TR due to the increase in SAR at the increased resonance frequency, since power is proportional to the square of the frequency. Higher fields also benefit from increased spectral dispersion (less overlap between peaks), so the longer TR is also outweighed by the improved metabolite detection.

Hope this helps!
Erin

2 Likes

Hi Erin,

Thank you for this excellent post - I agree that ‘canonized’ maybe wasn’t the word I should have used. It would be interesting to see a histogram of the TR used in MRS studies!

Yes, I’m very aware of FiTAID - currently looking into creating a similar, more generalized fitting framework. I should probably reach out to Roland!

Have a great weekend, and best wishes,
Georg

Hi,

just a short note on the TR in preclinical studies. we mainly use a long one 4-5sec.
best
cristina