Erroneous brain extraction using fsl and dwibiascorrect

Many of our subjects had parts of their frontal lobes removed during the brain extraction- we used FSL. What could be the reason? Is there a way to define a threshold to better extract the brain?

dwibiascorrect fsl ${SUB}_dwi_den_preproc.mif ${SUB}_dwi_den_preproc_unbiased.mif -bias ${SUB}_bias.mif

Thank you,

Hi Inbar,

Just to clarify, when you say ‘during the brain extraction’, you’re referring to the masking operation that happens internally within dwibiascorrect, right?

As you’ll have noticed when running this algorithm, there’s a warning shown on the terminal:

dwibiascorrect: [WARNING] Use of fsl algorithm in dwibiascorrect script is discouraged due to its strong dependence on brain masking (specifically its inability to correct voxels outside of this mask). Use of the ants algorithm is recommended for quantitative DWI analyses.

You’ll find more details about why the fsl approach is problematic in this post.

If you need to use the fsl approach regardless, I think one way around the issue might be to provide your own mask to dwibiascorrect using the -mask option, which sidesteps the problematic internal mask estimation procedure within dwibiascorrect.

1 Like