While the vast majority of MRS applications focus on the strong resonances of NAA, Cr, Cho and sometimes mIns and Glu+Gln, resonances from at least 15 neurochemicals, i.e., a comprehensive neurochemical profile are present in the spectrum. For detecting the small, weakly represented neurochemical resonances that underlie the typically detected large resonances such as NAA, Cr, Cho and mIns, options are: 1) to de-convolve all of the signals that are present or 2) to edit, i.e., to set the signal of interest apart (at least partially) from the others. Of course, there are advantages and disadvantages to each approach.
While the vast majority of MRS applications focus on the strong resonances of NAA, Cr, Cho and sometimes mIns and Glu+Gln, resonances from at least 15 neurochemicals, i.e., a comprehensive neurochemical profile are present in the spectrum [1, 2]. For detecting the small, weakly represented neurochemical resonances that underlie the typically detected large resonances such as NAA, Cr, Cho and mIns, options are: 1) to de-convolve all of the signals that are present or 2) to edit, i.e., to set the signal of interest apart (at least partially) from the others. Of course, there are advantages and disadvantages to each approach.
For the non-edited approach, detecting at ultra-high field (UHF) accommodates expansion of the neurochemical profile by making the signals stronger and more spread out (i.e., resolved), and thus easier to detect and de-convolve [3, 4]. It is important to remember radiofrequency (RF) excitation challenges scale with the magnetic field. That is, the same phenomenon that causes the signals to be spread out at higher fields also leads toward increased chemical shift dispersion if the rF excitation is not strong enough to overcome this dispersion [5]. Chemical shift displacement is a challenge that crosses into editing applications as well. An obvious advantage of non-edited detection is that all of the neurochemicals can be seen from one acquisition period. As such, one is able to gain information about more aspects of the biological or clinical mechanism of interest. For example, a focus of my work has been on oxidative stress, given pioneering efforts in the detection of the antioxidant GSH [6] and discovery of the resonance from vitamin C (Ascorbate, Asc [7]). Another advantage of non-edited approaches is that they can be implemented at short and ultra-short echo times. This becomes important when studying differences between biological or clinical states that might cause differences in the transverse relaxation rates, T2, for example in aging [8]. This is a problem because without making the scan prohibitively long (especially for clinical populations), the effect measured will be a mixture of concentration difference and T2 difference (and, actually other things), when the difference of interest is usually concentration. Practically, when attempting to measure the small concentrations differences that are of interest in many applications, the influence of T2 tends to be comparable in magnitude to that of the influence of concentration, even at relatively short echo time. It is for this reason that I recommend ultra-short echo time approaches to study aging.
A critical aspect of measuring neurochemical concentrations from a non-edited spectrum is the de-convolution or fitting procedure. Unwanted influences that tend to be minor relative to large signals such as NAA become major relative to the weakly represented signals such as GSH, GABA and Ascorbate (Asc). Therefore it becomes imperative that the basis set of the model is accurate, i.e., that it is measured appropriately or that the simulation accounts for all of the elements of the pulse sequence, including offset influences such as chemical shift displacement. It is also imperative that the broad, underlying macromolecule signal is characterized appropriately [9] [poster 2695], and that the spline baseline function is controlled effectively. Also, it is possible that inherent properties such as line width impact the quantification outcomes [[10], poster 2017]. It is imperative to understand correlation among measured concentration and the factors that can influence this. There are examples in the literature of approaches that have been used to show that outcomes of quantification of in vivo spectra reflect actual changes in concentration of the associated neurochemical [11] (and references therein) and to look at reproducibility [12, 13]. Finally, it is also worthy of mention that for a few neurochemicals, the spectral pattern that results from the coupling system makes detection at higher field less favorable, which we have found to be the case for glucose [13]. Also T2 tend to be shorter at higher field, so lines are broader, working against the benefit of higher spectral dispersion.
The term editing typically refers to use of scaler coupling between spins to selectively measure the desired molecule to the exclusion of the others. Homonuclear editing will be described from the perspective of the highly used MEGA-PRESS difference editing technique [14], although there are many excellent descriptions of this topic that can be referred to [15, 16]. Further, while some editing techniques such as multiple-quantum techniques require use of more complex theory, single-quantum coherence phenomena will be viewed using the classical model. This is adequate to illustrate how editing leads to the observable magnetization of commonly detected molecules such as GSH and GABA via MEGA-PRESS. Whereas chemical shifts give information about the environment of the nuclei (and are commonly viewed from the perspective of the ppm scale), scalar coupling (or spin-spin coupling or J-coupling) causes splitting of the resonance lines [17]. This splitting is caused by the fact that the resonance frequency of a given nucleus is influenced by the spin state of the nucleus that it is coupled to. Conceptually, editing inverts the spin state of a nucleus and impacts the phase evolution of the coupled nucleus. This is ultimately exploited to observe only magnetization that has this particular coupling system. Editing simplifies the spectrum because coupling systems are more unique that chemical shifts. However, GABA is a good exemplar for what happens when coupling systems are not sufficiently unique to avoid co-editing. Because J-coupling is constant over magnetic field, the editing pulses become more selective for a given molecule at higher field, a substantial advantage of UHF.
Of course, there are several practical matters to consider. When designing an editing approach, a key early step is to look for other coupling systems that could co-edit, particularly if the chemical shift of the signal of interest overlaps with that of a nuisance signal. An important exemplar is the overlap between GABA and macromolecules (MM). In order to minimize co-editing, a narrow bandwith editing pulse is desirable. However, optimization is constrained by the J-coupling constant, so the length of the editing pulse is accordingly constrained. Practically, editing pulses are not as narrow as would be desired, as exemplified by the fact that an editing pulse set at 1.9 ppm to edit for GABA has substantial excitation at 1.7 ppm, the chemical shift of a coupling partner for MM [18]. Accurate setting of the frequency of the editing pulse is also crucial given that the narrow bandwidth editing pulse profiles tend to be steep across the maximum. If the frequency of the editing pulse is off by as little as 10 Hz, the efficiency of editing can be altered. Thus the signal strength quantified would be not only a function of concentration but also the miss-setting of the frequency of the editing pulse. Frequency drift or change during an experiment can be particularly problematic to edited experiments. It is not unusual for motion of a subject’s head to cause a frequency drift up to 10 Hz, thus impacting editing efficiency. Worse, for co-edited signals such as GABA and MM, frequency drift causes the relative contributions of the desired (i.e., GABA) and undesirable (i.e., MM) signals. It is important to think about whether the cohorts in a study could differ with regards to how much they move in the scanner, because this could introduce bias. Approaches to minimize MM contamination are worthy of mention [19], although the importance of frequency-locking or at least monitoring frequency during the experiment still needs to be attended to. Finally, while the timing of the sequence and the offset of the editing pulse that gives optimal signal is determined by the coupling system, suboptimal designs still produce an edited signal and can be suitable under special circumstances [7, 20].
Quantification of the edited signal is simpler than that for non-edited signals. However it is not trivial and several approaches have been implemented [21, 22]. Note that with MEGA-PRESS one can use the summed spectrum to quantify the non-edited resonances, although the echo time will be relatively long. Taking the concept of editing more generally, there are other approaches that simplify an NMR spectrum to preferentially detect specific molecules. Longer TE spectra tend not to contain the MM signals that complicate quantification, so some prefer this approach. The fast T1 of MM can be used to edit for them using an inversion recovery approach [18]. The timing within STEAM can be optimized to accommodate resolving glutamate from glutamine [23].
2D NMR uses J-coupling to separate resonances along orthogonal axes [24]. For COSY, the general idea is to generate a second frequency axis by introducing an evolution delay and a pulse into a 90 degree – acquire experiment. To use COSY in vivo, the basic sequence is extended to allow spatial localization. J-resolved NMR resolves scalar coupling fine structure of resonances and their chemical shifts in two orthogonal directions. This is basically a spin-echo experiment in which the evolution period is the echo time. It can be readily performed in vivo using a double spin echo such as PRESS.
Bogner et al [25] is a publically available explanation of 2D NMR. Note that it is has a comparable description of J-difference editing. Intramolecular spin-spin coupling is used to create 2D spectra characterized by different frequencies along orthogonal spectral dimensions. The distance between resonances in the 2D plane is greater than in 1D, increasing resolution. A disadvantage is lower SNR. The acquisition time tends to be long in vivo and data quality is susceptible to instabilities. ProFit [26] has been developed for quantification but it only works for concentrations normalized to creatine, i.e., not absolute quantification.
1. Govindaraju, V., K. Young, and A.A. Maudsley, Proton NMR chemical shifts and coupling constants for brain metabolites. NMR Biomed, 2000. 13(3): p. 129-53.
2. Pfeuffer, J., et al., Toward an in vivo neurochemical profile: quantification of 18 metabolites in short-echo-time (1)H NMR spectra of the rat brain. J Magn Reson, 1999. 141(1): p. 104-20.
3. Emir, U.E., et al., Regional neurochemical profiles in the human brain measured by (1)H MRS at 7 T using local B(1) shimming. NMR Biomed, 2012. 25(1): p. 152-60.
4. Deelchand, D.K., I. Iltis, and P.G. Henry, Improved quantification precision of human brain short echo-time (1) H magnetic resonance spectroscopy at high magnetic field: a simulation study. Magn Reson Med, 2014. 72(1): p. 20-5.
5. Chadzynski, G.L., et al., In vivo proton magnetic resonance spectroscopic imaging of the healthy human brain at 9.4 T: initial experience. MAGMA, 2015. 28(3): p. 239-49.
6. Terpstra, M., P.G. Henry, and R. Gruetter, Measurement of reduced glutathione (GSH) in human brain using LCModel analysis of difference-edited spectra. Magn Reson Med, 2003. 50(1): p. 19-23.
7. Terpstra, M. and R. Gruetter, 1H NMR detection of vitamin C in human brain in vivo. Magn Reson Med, 2004. 51(2): p. 225-9.
8. Marjanska, M., et al., Faster metabolite (1)H transverse relaxation in the elder human brain. PLoS One, 2013. 8(10): p. e77572.
9. Cudalbu, C., V. Mlynarik, and R. Gruetter, Handling macromolecule signals in the quantification of the neurochemical profile. J Alzheimers Dis, 2012. 31 Suppl 3: p. S101-15.
10. Marshall, I., et al., Use of Voigt lineshape for quantification of in vivo 1H spectra. Magn Reson Med, 1997. 37(5): p. 651-7.
11. Deelchand, D.K., et al., Sensitivity and specificity of human brain glutathione concentrations measured using short-TE (1) H MRS at 7 T. NMR Biomed, 2016. 29(5): p. 600-6.
12. Prinsen, H., et al., Reproducibility measurement of glutathione, GABA, and glutamate: Towards in vivo neurochemical profiling of multiple sclerosis with MR spectroscopy at 7T. J Magn Reson Imaging, 2017. 45(1): p. 187-198.
13. Terpstra, M., et al., Test-retest reproducibility of neurochemical profiles with short-echo, single-voxel MR spectroscopy at 3T and 7T. Magn Reson Med, 2016. 76(4): p. 1083-91.
14. Mescher, M., et al., Simultaneous in vivo spectral editing and water suppression. NMR Biomed, 1998. 11(6): p. 266-72.
15. De Graaf, R.A., Spectral Editing and Two-dimensional NMR, in In Vivo NMR Spectroscopy: Principles and Techniques. 2007, Jonh Wiley & Sons Ltd.: West Sussex, England. p. 389-440.
16. Bendall, M.R., et al., NMR of Protons Coupled to 13C Nuclei Only. J Am Chem Soc, 1981. 103(4): p. 934-936.
17. Ramsey, N.F. and E.M. Purcell, Interactions between nuclear spins in molecules. Phys Rev, 1952. 85: p. 143-144.
18. Behar, K.L., et al., Analysis of macromolecule resonances in 1H NMR spectra of human brain. Magn Reson Med, 1994. 32(3): p. 294-302.
19. Henry, P.G., et al., Brain GABA editing without macromolecule contamination. Magn Reson Med, 2001. 45(3): p. 517-20.
20. Oz, G., et al., Proton MRS of the unilateral substantia nigra in the human brain at 4 tesla: detection of high GABA concentrations. Magn Reson Med, 2006. 55(2): p. 296-301.
21. Terpstra, M., et al., Detection of an antioxidant profile in the human brain in vivo via double editing with MEGA-PRESS. Magn Reson Med, 2006. 56(6): p. 1192-9.
22. Edden, R.A., et al., Gannet: A batch-processing tool for the quantitative analysis of gamma-aminobutyric acid-edited MR spectroscopy spectra. J Magn Reson Imaging, 2014. 40(6): p. 1445-52.
23. Yang, S., et al., Spectral simplification for resolved glutamate and glutamine measurement using a standard STEAM sequence with optimized timing parameters at 3, 4, 4.7, 7, and 9.4T. Magn Reson Med, 2008. 59(2): p. 236-44.
24. De Graaf, R.A., Spectral editing and two-dimensional NMR, in in vivo NMR spectroscopy. 2007, John Wiley & Sons Ltd: West Sussex. p. 389-440.
25. Bogner, W., et al., 1D-spectral editing and 2D multispectral in vivo1H-MRS and 1H-MRSI - Methods and applications. Anal Biochem, 2016.
26. Fuchs, A., et al., ProFit revisited. Magn Reson Med, 2014. 71(2): p. 458-68.