[RESOLVED] T2 FLAIR prefreesurfer pipeline

Feature Requests
1

Hi everyone

We are working with non HCP data, that has a T2 FLAIR, that we are trying to process through the HCP data pipeline. The issue we have is that the pre-freesurfer BrainExtraction_FNIRT performs really badly when creating the mask and extracting the T2 flair image and affects the rest of the pipeline.

The problem seems to be the T2 MNI templates in HCP pipeline used for registration doesn’t include a FLAIR option and the HCP pipeline doesn’t register the T2 to the T1 before extraction. We have found we can get good brain extraction when we register the T2 FLAIR to the T1 before running it through the BrainExtraction_FNIRTbased.sh script.

Would it be possible to include in the pipeline a FLAIR T2 MNI template to be used in the pipeline? Or alternatively, an option to change the brain extraction in BrainExtraction_FNIRT.sh to include registering the T2 FLAIR to the T1? This would be useful as it would allow the HCP pipeline to be run on data with T2 FLAIRs.

I am happy to share the T1 and T2 FLAIR we used as well as a more detailed breakdown of what we did if helpful.

Best wishes
Daniel

2

Hi Daniel,

Thanks for this suggestion. As this upgrade requires both QuNex and HCP pipelines changes, I will run this by our HCP colleagues and see what they think. I will keep you posted here.

Best, Jure

3

Hi Daniel,

I talked with HCP folks. They said that they do not have enough T2 FLAIR data to create such template.

Furthermore, they said that (so far) they did not experience issues you are describing when using T2 FLAIR, but they could be lucky because they did not use this extensively.

Are you using a 3D hires T2 FLAIR? One thing you could try is to use the T1w image instead of the T2w image for the templates and see if that helps. If you have enough data that works, you could make your own T2 FLAIR template for the rest that doesn’t work.

Best, Jure

4

Hi Jure,

Thanks for getting back to me and your suggestions. Yes it is a hires T2 FLAIR but I think we might try and use another template like the MIITRA templates.

Best wishes
Daniel