Skip to content

zalteck/TV-Deep-SR-fMRI

Repository files navigation

TV-BASED DEEP 3D SELF SUPER-RESOLUTION FOR FMRI

DL Self Super Resolution

This repository implements the code for the paper submitted to the International Symposium on Biomedical Imaging (ISBI) 2025.

The preprint can be found in Arxiv

https://arxiv.org/abs/2410.04097

Image

AFNI commands for preprocessing:

Mask:

>3dAutomask -prefix sub_mask.nii.gz input.nii.gz

Preprocessing:

>3dcalc -overwrite -a 'input.nii.gz[5..$]' -expr 'a' -prefix sub_tcat.nii.gz
>3dvolreg -overwrite -verbose -Fourier -prefix sub_volreg.nii.gz -base 0 -1Dfile sub_motion.1D -1Dmatrix_save sub_motion -maxdisp1D sub_maxdisp.1D sub_tcat.nii.gz
>1d_tool.py -infile sub_motion.1D -censor_motion 0.3 sub
>1d_tool.py -infile sub_motion.1D -derivative -demean -write sub_motion.deriv.1D
>1d_tool.py -infile sub_motion.1D -demean -write sub_motion.demean.1D
>3dTproject -overwrite -input sub_volreg.nii.gz -prefix sub_tproject.nii.gz -polort 4 -censor sub_censor_censor.1D -cenmode KILL -ort sub_motion.demean.1D -ort sub_motion.deriv.1D -passband 0.005 0.2 -mask sub_mask.nii.gz

Correlation maps are calculated using AFNI InstaCorr with:
Blur = 3mm, seed rad = 3

About

DL Self Super Resolution for fMRI

Resources

Stars

Watchers

Forks

Packages

No packages published

Languages