-
Notifications
You must be signed in to change notification settings - Fork 1
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Detect mis-alignment #6
Comments
@giulia-berto Can you think of a way to quantify how "odd" the alignment is? |
shall we train a neural network? |
Sure! I am all open to your idea! This has to work on pretty much any resources and with as light dependencies as possible.. |
Another easier way to show to the user that images are not aligned properly might be to show what the image should look like by underlying MNI space t1 image so user can contract it. Also, can we add L/R A/P S/I labels on the output image? |
Is there some relationship here with whether the background has been masked out or not? If the background has not been masked out, then there will be both background noise and zero entries in the datablock. In the case of a weird alignment in a non-background masked nifti, the border between the zero (empty) entries and the background would be non-orthogonal in the data-block? In these weird cases is the off-diagonal of the affine all zeros? There could be several issues compounding these outcomes. |
Maybe the misalignment is caused by the non-masked out background. Regardless of the reasons for misalignment, though, we should have a simple way of alerting users about the "potential" misalignment. |
I don't know if something has been done yet to warn the user of the misalignment. I forgot about it, and now it's not the best moment for me to work on that. But I agree that it would be nice to have a simple way of doing that. |
Often "alignment" app fails to align data
From neuro/anat validator point of view, data aligned in any orientation is still valid, but I think it would be nice if we could somehow detect when the image looks "odd" and give warnings to the user - some novice user might not even know that the image isn't supposed to look like this after "ACPC Alignment".
The text was updated successfully, but these errors were encountered: