Rapidtide seems to remove part of the brain

[Annen]

Hi,

I am using rapidtide, and on most of my fMRI data it works well. However, for some sessions of a single subject it seems to cut off part of the brain:

This is the input data:

And after using rapidtide, the data looks like this:

Strange thing is that for other sessions of the same subject, the data looks much better:

Is this a known issue? And how could I solve it? It would be a pity to exclude subjects from the study.

Thanks for your help

[bbfrederick]

What's happening here is that the automatic mask creation is not working right, because the range of mean intensities of the data is not what rapidtide is expecting (the mean intensity is fading dramatically towards the back of the brain).

The best solution for this is if you have a brain mask that you think is correct, you can tell rapidtide about it with the

--brainmask FILENAME

option. That will use the mask you give it everywhere a brain mask is needed rather than trying to estimate one.

If you don't have a mask and don't want to create one (but you probably should), you can try setting --corrmaskthreshpct PCT to something lower than its default of 1.0. That may bring in some voxels from outside the brain, but it should bring back the back of the head.

[Annen]

I have tried both options on different subjects.
First, I used the explicit mask, however, this did not have the effect I expected, as still there are voxels excluded (actually the same as without specification of the mask).

In grey is the rapidtide cleaned data, in yellow the original data, and red is the mask I used.

Then I tried the option --corrmaskthreshpct (once with 0.8, and once with 0.6) on another subject, and again, it does not change the output file it seems.

I suppose I must be doing something wrong, so I hope you can help me further in solving this issue.

[bbfrederick]

The two datasets you're showing here are different, right? The distortions are not the same. I'm not sure how to interpret the top image - it looks like most of the brain is included. The dropout above the roof of the mouth looks like a standard susceptibility artifact - I'm not sure anything would process signal in that area.

[Annen]

Yes indeed, the two datasets are different. I tried the solution on several.
Perhaps you are right for the first dataset, and I could eventually include it. However, in the case of the dataset of my first post, a more serious proportion of the cortex is missing. Do you see any other workaround than the two solutions that you proposed but do not have the anticipated effect?