longitudinal option for T1 processes about tractoflow

[arnaudbore]

Seems to be the way to go:
https://github.com/ANTsX/ANTs/blob/master/Scripts/antsMultivariateTemplateConstruction2.sh

[cbedetti]

I was thinking only for T1 but a multivariate version seems even better.

The template construction process should be done just before Resample DWI and Resample T1, right ?

But at this stage of the pipeline :

• T1s are not yet brain extracted or cropped
• DTI metrics are not produced

So the Resample T1 process should be delayed and temporary DTI metrics should be computed after the Normalize DWI process.

Is that right ?

[jchoude]

1. I don't think we need the multivariate in the sense of using DTI metrics as well. @frheault has experience in building templates of T1s.

2. What would be the goal, here? How would it influence the rest of the pipeline?

My main concern is that this is quite a long process, and would potentially (depending on which step is planned to use that template) block multiple other processes, which we normally avoid as much as possible.

Depending on the goal of the template building step, it might better be suited as a separate workflow. Our (well, at least mine) philosophy is that writing separate workflows for tasks that do not directly impact the dMRI processing steps is often better, since it allows to better distribute compute resources, etc.