QSM_SSTV pipeline ... What's wrong here?

, ,
#1

Hi Everyone! I was hoping that you could help me out with something that has been bothering me for a couple of days already :frowning:

First of all, I followed instructions from https://github.com/AlanKuurstra/qsm_sstv/blob/master/run_instructions.md to create folders, retrieve data from the dicom server, and then to run the QSM pipeline. From figure 1, you can see that I have downloaded some subjects already, and for the sake of the explanation, I will focus on the data generated for subject 7 (sub-7).

image
image926×342 12.7 KB

Then, I configured the QSMBatch.sh file to start executing the pipeline. I filled the file as in figure 2.

image

However, when I ran the job on compute Canada, I don’t find the qsm result. Instead, I see an output with what you can see in figure 3.

image

Any ideas of what is happening? thanks in advance,
Have a nice day :smiley:

#2

I don’t see any relevant images in your BIDS anat folder (e.g. magnitude and phase GRE images) – what did you use to convert this to BIDS?

#3

from /home/gvarelam/, I used the following command:

“”" cfmm2tar -d ‘20210224’ -p ‘Menon^QSM’ -n ‘2021_02_24_subject_7’ /home/gvarelam/scratch/QSM/tar “”".

and then on /home/gvarelam/scratch/QSM, I used:

“”" singularity run -B tar:/input -B bids:/output /home/gvarelam/tar2bids_latest.sif -o /output /input/Menon_QSM_20210224_2021_02_24_subject_7_1.82C6632C.tar “”"

NOTE: In that session we did several ME-GRE with different accelerations factors from 1x to 16x (1x: 1x1 ; 4x: 2x2 ; 9x: 3x3; 16x: 4x4).

#4

I’m guessing the default heuristic in tar2bids isn’t picking up your GRE scans.

Here you can see how it finds them:
https://github.com/khanlab/tar2bids/blob/0646e9a91da2d85f51805e65f11b8e022a2fc91a/heuristics/cfmm_base.py#L489

It checks whether susc, gre3d or t1_fl3d_p4_iso is in the series description.

 if ('susc' in s.series_description or 'gre3d' in s.series_description or 't1_fl3d_p4_iso' in s.series_description ):

But the other issue is the heuristic is only set-up for a single GRE scan…

So I would suggest you edit this cfmm_base.py (or create a new one) to add in your different acceleration factors, so they each get mapped to a different file… @switt4 could assist with this if needed. Or do the bids conversion in another way (manual, or another tool).

#5

Actually, if you tell me what series_description you have for these scans, I can add run-XX to the bids naming pretty easily.

#6

Hi @akhanf, thanks for your help on this one! The series description, if I get it right, would be gre_CS_C41_iso_FS, gre_CS_C41_iso_P4, … P9, …P1, and so on. Therefore, gre_CS_C41* should do the trick for all of them :d.

image
image951×454 54.2 KB

#7

I’ve made a new version here https://github.com/khanlab/tar2bids/pull/19 but will need you to test it out.

You could either download the heuristic from github now, and use tar2bids -h <path_to_downloaded_heuristic.py> ... when converting,

or,

if you wait for the CI to complete (maybe 15 minutes) there will be a docker://khanlab/tar2bids:cs-gre container you can download and try.

#8

Hi @akhanf,

Btw, can we also include the following keywords in the heuristic:

Aspire_M*, Aspire_P_, R2star_ASPIRE, T2star_ASPIRE*.

ASPIRE is a new sequence, at CFMM, with a novel reconstruction procedure that provides an improved coil phase combination for phase images. This is ideal for QSM.

image
image874×326 53.8 KB

#9

Thanks – actually was just about to do this, as acquired it in a patient today!
Images look great!

1 Like