The document summarizes a student's project using computational spectroscopy techniques to analyze brain metabolites like gamma-aminobutyric acid (GABA). The student used non-uniform sampling and maximum entropy reconstruction to obtain high-resolution spectra from sparse datasets. Parameters like line width and J-coupling constants were varied. Integrated peak areas from the reconstructed spectra were plotted against the parameters. Further analysis of molecular structures was done using NMR simulation software. The pulse program was modified to better analyze relaxation times and magnetic properties. The student learned about NMR and MRI techniques and hopes to continue the project.

1. Shaunak Mukherjee,
Centre for Advanced Imaging,
University of Queensland.
1CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

2. 2-D COMPUTATIONAL
SPECTROSCOPY OF BRAIN
METABOLITES
A COMPUTER SCIENCE STUDENT’S HUMBLE
SOJOURN INTO UNCHARTERED WATERS
2CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

3. The (initial) objectives
Using non-uniform sampling (NUS) to obtain high-resolution spectra from relatively sparse
data records, from compounds such as gamma-aminobutyric acid (GABA)
Using the Maximum entropy (MaxEnt) reconstruction principle, (a non-FT method of spectrum
analysis) to effectively analyze the high-resolution spectra thus obtained from the NUS data set
Changing the parameters governing them and observing the effect on the aforementioned
spectra
3CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

4. So how did we start?
Well, the first step was to obtain the data sets of those spectra, more specifically, the “cosynus”
and the “nusidx” files.
We then tweaked the two parameters :
A) the line-width of the spectra (“LW”)
B) the J-coupling constant between the various protons in the GABA molecule
4CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

5. And how did we do that?
5CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

6. The picture represents the compilation of the data file
(containing the changed parameters) through the
/proc2.com command in the ASAP workstation.
6CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

7. The result of the compilation is obviously, a Maximum Entropy Reconstructed file (me_recon_ser.ft2).
BUT WHAT CAN WE POSSIBLY DO WITH THAT?
As we know, the .ft2 file will not really come of use to us in order to analyze the spectra on-screen.
WE CONVERT IT TO A .UCSF FILE.
As Sparky itself puts it, -
“A UCSF NMR data file is a binary file with a 180 byte header, followed by 128 byte headers for each axis of the spectrum,
followed by the spectrum data. All integer and float values in the file have big endian byte order independent of the native
byte order for the machine that wrote the file”.
7CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

8. So how do we do that?
By adding the “pipe2ucsf” command.
pipe2ucsf me_recon_ser.ft2 lw<#>J<#>.ucsf
Where, ‘#’ denotes the value of the line-width and the J-coupling.
Now that it has finally been converted to the .UCSF, we go to Sparky, and run it, where we get
the following spectra.
NOTE : Only the cross and diagonal peaks pertaining to the observed molecules have been
displayed.
8CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

9. The spectra obtained in the Sparky
interface.
9CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

10. Yes, yes. The pattern looks beautiful.
What do we do with it, then?
We now use the various functions in Sparky, and integrate the area under the various peaks, and
build a specific data set with the line-widths, the J-coupling, and the areas –incrementing the
“lw” and “J” by 2 units each time, finally obtaining a grand total of 363 data entries.
Sample data-set :
LW18.0
j0
1.869 2.976 lw 26.632 28.704 vol 2.481e+05 rms 16.2%
1.870 2.258 lw 22.200 25.311 vol 2.061e+05 rms 21.5%
1.869 1.869 lw 33.795 41.196 vol 5.540e+05 rms 25.9%
j2
1.870 2.976 lw 26.515 28.600 vol 2.470e+05 rms 14.7%
1.870 2.259 lw 22.637 25.191 vol 2.073e+05 rms 18.0%
1.869 1.869 lw 33.918 41.057 vol 5.566e+05 rms 21.9%
10CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

11. And now..
We plot the graphs – between the “lw”, the “j-coupling” and the areas of the peaks.
0
100
200
300
400
500
600
700
800
0 5 10 15 20 25
LW V/S (AREA OF) P1
0
50000
100000
150000
200000
250000
300000
350000
400000
450000
0 5 10 15 20 25
LW V/S (AREA OF) P2
11CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

12. On to the next phase.
We now delve into the actual structure of the GABA and the Creatine molecules, and try to
analyze them using certain other paramaters.
12CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

13. Further analysis?
We use TOPSPIN, another software by Bruker, to analyze the data further, specifically using the
NMR simulation tool.
13CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

14. The hamiltonian i.e. The spin system, configured with
GABA, creatine, NAA, etc.
14CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

15. The original pulse program -
15CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

16. 16CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

17. It needs modifying, doesn’t it?
17CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

18. So what did we do there?
Modified the pulses –
> added certain new waveforms – “Gauss1.1000”, “sinc1.1000”, “uqcai.rf”
>added some new delays
>modified the phases of the pulses
18CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

19. Important parameters
P4 : 42.8 Hz
P1, P2 : 831.7 Hz
SPNAM :
i) sinc1.1000
Ii) Gauss1.1000
Iii) uqcai.rf
DELAYS : d1 = 1s, d2 = d3 = d4 = d5 = D
4D + P1 + 2P2 + 2P4 = (23 + 4*26.5) ms = 129 ms. ( 1 / J ) -> echo time
19CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

20. Question is, WHY?
> Modifying the echo time ( to be 1/ J where the coupling constant is J ) , relaxation time,
acquisition time, etc.
WHY?
>Let’s delve into the magnetic side of things.
>The next slide will make it clearer.
20CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

21. Courtesy : Wikipedia
21CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

22. The given Bloch module – (The pulse is
originally applied as Mz)
22CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

23. We tweak the power of the pulses given in increments of 2 Hz,
and plot the differences in the integrated areas of the peaks
obtained in the GABA and Creatine molecules. Enclosed, is the
plot.
23CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

24. SO WHAT DID WE LEARN?
A whole lot.
Forayed into NMR, MRI and all other difficult abbreviations, and kind of enjoyed it.
FUTURE WORK?
If I have the pleasure of coming back here again, will try to continue the data processing and
obtain even better spectra, and analyze them.
24CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

25. Acknowledgments.
Thank YOU, Dr. Tesiram. Wouldn’t have been possible without you.
Thank you to, well, so many people – Aish, Theo, Hari, Gagan, Rajiv, Sami, Venkat – basicallt
everyone at CAI. This project would have drowned in the beach without you guys.
Thank you for making this the best summer ever. :’)
25CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015

26. THAT’S ALL, FOLKS. :D
26CENTRE FOR ADVANCED IMAGING, UNIVERSITY OF QUEENSLAND27-07-2015