Adapting the Synb0 disco pipeline for Feline data

[archithrajan711]

Hi Experts,

I have some feline data which is heavily distorted in the L-R direction and without a reverse b0 acquisition, on which I'd like to run the pipeline.

1. To start off with, would this be even feasible considering the model was trained on human brain data?

2. If there's a chance, what modifications would one need to perform, especially in the steps where non-linear registrations to MNI templates have been performed?I do have a few Feline templates if that could be of any use.

Regards,
Archith

[hanscol]

Hello Archith,

Unfortunately, we have no reason to believe the model will generalize to other species, though we have not tried. If you still want to run the model on your data, you could run the docker interactively with bash as your entrypoint and then follow the steps in our pipeline as close as possible.

Best,
Colin

[schillkg]

Hi Archith,

Like Colin said - I'm aren't sure if this will generalize to non human data, but it is worth a shot! If you do give it a run, I would be happy to walk through the results and we can debug where the algorithm may first go wrong (my guess is that it might have a hard time synthesizing the non-distorted data). Regardless, I'm excited to see what it does. Modifications may be needed in the registration to standard space, and maybe a contrast normalization (to make a structural image look "T1w"), as well as manipulation of TOPUP configuration file, but it is still possible it may work right out of the box.

Thank you,
Kurt

[archithrajan711]

Dear Colin and Kurt,
Thank you for the replies.I'll give it a try and get back soon.

Warm Regards,
Archith