search

MASILab / Synb0-DISCO

Distortion correction of diffusion weighted MRI without reverse phase-encoding scans or field-maps
https://my.vanderbilt.edu/masi
58 stars 29 forks source link

Difference in output between 3.0 and previous version #34

Closed robbwh closed 2 years ago

robbwh commented 2 years ago

Hi from a VMAC PhD student,

I just wanted to point out a difference I noticed between the previous and new docker image: hansencb/synb0:latest leonyichencai/synb0-disco:latest

I am using this tool, unconventionally, to correct for distortion in BOLD fMRI images. So no need to troubleshoot for my sake.

New version b0_u.nii.gz: Darker than expected ventricles (maybe due to spreading out the choroid plexus values)

Screen Shot 2022-09-13 at 2 56 31 PM

Older version b0_u.nii.gz doesn't spread out choroid plexus values

Screen Shot 2022-09-13 at 2 57 41 PM

I used the --stripped command for the new version and input the T1_mask.nii.gz and T1.nii.gz for the old version, which I thought would give similar performance.

Not sure why this might be, but thought I might as well point it out.

leonyichencai commented 2 years ago

Hello! For the old version, was T1.nii.gz already skull stripped? Same for T1_mask.nii.gz? The registration step changed slightly to more accurately account for stripped inputs between hansencb/synb0:latest and leonyichencai/synb0-disco:latest. But also, we're happy to point you to this repo where we have implemented Synb0 for fMRI BOLD (called SynBOLD)! We are still in the validation stages, but would love to have your input: https://github.com/MASILab/SynBOLD-DisCo

robbwh commented 2 years ago

Thanks for pointing me to that repo--will check it out shortly!

For the old version: T1.nii.gz was not skull stripped but T1_mask.nii.gz was.

I'll see what happens using the BOLD version.

robbwh commented 2 years ago

Tried out SynBOLD-DisCo and choroid plexus looks good now--there is another slight issue with the first person I tried, but I will make an issue on that github page