I’m getting negative GABA values…
This was during a visual task in the visual cortex:
Has anyone else seen something like this?
Has anyone else seen something like this?
I think this is likely an error with the water being positive
In commonest errors Examples — Gannet
It shows a similar example
And seems to suggest I can change the MRS_struct to include:
ONOFFfirst = ‘offfirst’ or ‘onfirst’
or
Waterpositive = ‘positive’ or ‘negative’
But changing MRS_struct.p.ON_OFF_first to ‘offfirst’ or ‘onfirst’ doesn’t seem to change anything
and I can’t find any ‘waterpositive’ field
Hi @weberam2,
In the latest version of Gannet, the phase of the data is determined automatically. The examples you linked to are for prior versions of Gannet.
In this case, the automatic phasing has failed. You will need to invert the spectra manually. The simplest way to do this is to add this line of code right after line 400 in GannetLoad.m:
MRS_struct.fids.data = -MRS_struct.fids.data;
Remember to comment this line out for all other datasets.
Mark
This worked! Thank YOU!
hi all,
I am getting a similar issue with negative GABA values.
I’m using Gannet 3.3.1, and so added in the code to invert the spectra manually, but now my Cr Frequency chart looks messed up. Did I do the right thing?
Original.pdf (526.5 KB)
SpectraFlipped.pdf (531.7 KB)
(It might be that this data is poor quality, and unusable, but I’m just trying to figure out the flips!)
thanks!
Hey @mmikkel
Thanks for a speedy response!
I threw it into GannetLoad.m right about line 400 - between 'Global zero-order phase correction" and “Line-broadening, zero-filling and FFT”.
Molly
Hmm, are you able to share your data so I can debug? mam4041@med.cornell.edu
of course! will send this afternoon - much appreciated!
1 Like
Hi，PROFESSOR，
i`m using the 3.3.2, and i have encounter the same question-“negative GABA”,
Then I try to put the code after the line 400 as this:
and i also got the Negative GABA quantification,
Can you help me to see if this is due to a data quality issue, thanks！
I would suggest changing alignment in GannetPreInitialise.m to 'SpecReg'. If that doesn’t help, set it to 'none'.
waooo, Thank you professor, it really works after turn ‘RobustSpecReg’ to ‘SpecReg’, and may i ask that could i use the same ‘GannetPreInitialise.m’ to analysis the same batch of participant.
Yes, use the same GannetPreInitialise.m settings for the rest of your participants.
Thank you for your response，professor. I’ve come to understand that SpecReg is better suited for datasets where the spectral peaks are already relatively aligned. It seems that SpecReg could be more concise and effective for data previously processed by the imaging department’s staff.
hello professor, happy new year，
i have used this GannetPreInitialise to analysis the other data, and some of my data also encountered the"negative gaba":
and the other data did not have this problem:
i have changed the " MRS_struct.p.alignment = ‘SpecReg’; ".
i want to ask that if I encounter a few such data points individually in my analysis, how should I make minor adjustments?
Hi @Gao,
I suggest you use the upcoming version of Gannet (v3.4.0), which can be downloaded on the development branch on GitHub: GitHub - markmikkelsen/Gannet at dev
In this version, open GannetPreInitialise.m and select either offfirst or onfirst for MRS_struct.p.ON_OFF_order. This should resolve the issue. You can still use RobustSpecReg in this instance.
Mark
thank you Mr mmikkel，
happy new year, i have download the gannet-dev, and also find negative GABA,
but when i set (MRS_struct.p.ON_OFF_order = ‘onfirst’
and ( MRS_struct.p.alignment = ‘none’), the GABA become positive.
when set ( MRS_struct.p.ON_OFF_order = ‘offfirst’) and (MRS_struct.p.alignment = ‘RobustSpecReg’) the result is still negative:
Hi @Gao, it’s likely that your data are not of good quality. Could you please share the GannetLoad outputs?
hello, professor mmikkel, it is seemed like that the data are not so good, because i re-download the new data from our imaging platform. I still use the v3.3.2 get the positive GABA.
Thank you for your reply!
Hi @mmikkel,
I have run 89 scans in Gannet as a batch; 10 of them have inverted spectra (example attached). When run individually, the inverted spectra are corrected when using:
MRS_struct.fids.data = -MRS_struct.fids.data;
However, I have noticed batch effects between batches of scans analysed separately using Osprey so I am keen to run all 89 scans as a single batch, rather than run the 10 negative spectra separately. Could I use an ifelse in GannetLoad.m to only apply the additional line of code to the 10 inverted scans?
Alternatively, if you do not think that running the same study in different batches in Gannet is an issue, then that would be reassuring, and I will just run them separately.
Many thanks,
James
