How to deal with intensity drift in T1w anatomicals?


I’m working on fmri data preprocessing, and I’m facing this issue with T1w anatomicals having an intensity drift. As can be seen on the attached image, inferior slices have lower intensities and this poses issues in the segmentation step because tissue won’t be recognized correctly at the bottom brain slices. I was wondering if someone could recommend a method/software that can help me homogenize the intensity in my T1w anatomical images.

Thank you!

Hi @CanisLupusLycaon ,

There are many options to deal with such inhomogeneities but in my experience N4 algorithm works quite well as an all-around solution.

You might start from 3DSlicer’s implementation:

1 Like


Another option would be to use AFNI’s 3dUnifize to stanardize+homogenize the brightness across the volume:
The program outputs an anatomical volume with WM scaled to an approx. uniform scaled value of 1000, and you can provide an option to try to scale the GM to an approximate uniform brightness, too. For example:

3dUnifize                             \
    -GM                               \
    -Urad 30                          \
    -prefix DSET_OUT                  \
    -input DSET_IN

(The “-Urad …” option sets a better radius for brightness for the size of adult human brains.)



Thank you all for your helpful advice. I ended up using SPM’s segmentation workflow which did a relatively good job on my 7T anatomicals. I’ll keep your recommendations in mind for the future.

1 Like