Removing background noise from MP2RAGE images

Tutorials
, ,
#1

Since a previous scanner software upgrade, the MP2RAGE sequence on the 7T does not produce the UNI-DEN (background-noise removed, T1w-like image) automatically. But you can generate it from the data that is acquired.

Here are a few options for this:

  1. Use tar2bids to a dicom tarball to bids, as it has this extra step built in. The docker container is at docker://khanlab/tar2bids:latest
  2. Use the standalone Python script that Option 1 has, which can be found here:
    https://github.com/khanlab/tar2bids/blob/master/etc/mp2rage_genUniDen.py
  3. Use the equivalent MATLAB script, located here:
    https://github.com/khanlab/mp2rage_genUniDen

Usage: python mp2rage_genUniDen.py MP2RAGE_filenameUNI MP2RAGE_filenameINV1 MP2RAGE_filenameINV2 MP2RAGE_uniden_output_filename [multiplyingFactor]

(default=6, increase up to 10 for more noise suppression)

adapted from RobustCombination function from Jose Marques, https://github.com/JosePMarques/MP2RAGE-related-scripts
this function shows one possible implementation of the methods suggested
in http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0099676

#2

Another option, as well as some additional MP2RAGE cleaning steps (e.g., removal of pads, skullstripping) can be found here: https://github.com/srikash/presurfer

Works pretty good!

#3

I’m trying to run the matlab version, and I am getting the following error:

>> mp2rage_genUniDen('sub-CW20_acq-UNI_run-01_MP2RAGE.nii.gz','sub-CW20_inv-1_run-01_MP2RAGE.nii.gz','sub-CW20_inv-2_run-01_MP2RAGE.nii.gz','output_8.nii.gz',8);
using multiplying factor of 8
Error using  .* 
Integers can only be combined with integers of the same class, or scalar
doubles.

Error in mp2rage_genUniDen (line 49)
INV1img.img=sign(MP2RAGEimg.img).*double(INV1img.img);

Any suggestions as to what might be going wrong?

Thanks,
suzanne

#4

Does it run successfully for you with the default multiplying factor? If so, then the problem may be elsewhere…

#5

I just tried running the matlab version using the default multiplying factor, and I get the same error. (I did get the python version from tar2bids up and running.)

#6

For those who may be switching over to using MP2RAGE on the CFMM 3T, a bit of trial and error suggests that using a multiplying factor of 300 produces effective background suppression.

There is also a new tar2bids container, tar2bids:uniden3T, that includes a new argument, -U, where one can specify a custom multiplying factor.

The default multiplying factor in tar2bids is set to 6, the optimal value for background suppression for the 7T MP2RAGE sequence.