Hi AFQ team!
We have been analysing several AFQ processed datasets, and we have a suggestion for how the clip2rois parameter is working.
Right now, for both the 'true' and 'false' options, the resulting tract profile is "normalized" to 100 divisions. The main issue we find with this is that the profiles (or averages) coming from two different clip2roi options cannot be compared, although both have the same number of divisions.
We think that it could be useful if the clip2roi=true option would have the 100 divisions, and the clip2roi=false option maybe 150 or 200 divisions. One way of getting rid completely of the parameter would be to assign 100 divisions between the ROIs always, and use 50 divisions from the ROIs to the GM in both ends. This way we always would have 200 divisions, and could easily select the middle 100 for the clip2roi segment of the tract profile (maybe another option for backward compatibility would be to maintain always the 100 divisions, but make sure that the middle 50-60 correspond to the clip2roi section of the tract profile).
What do you think? I can create a pull request if you think this change is worth implementing.
Thanks!
Gari
Hi AFQ team! We have been analysing several AFQ processed datasets, and we have a suggestion for how the clip2rois parameter is working. Right now, for both the 'true' and 'false' options, the resulting tract profile is "normalized" to 100 divisions. The main issue we find with this is that the profiles (or averages) coming from two different clip2roi options cannot be compared, although both have the same number of divisions.
We think that it could be useful if the clip2roi=true option would have the 100 divisions, and the clip2roi=false option maybe 150 or 200 divisions. One way of getting rid completely of the parameter would be to assign 100 divisions between the ROIs always, and use 50 divisions from the ROIs to the GM in both ends. This way we always would have 200 divisions, and could easily select the middle 100 for the clip2roi segment of the tract profile (maybe another option for backward compatibility would be to maintain always the 100 divisions, but make sure that the middle 50-60 correspond to the clip2roi section of the tract profile).
What do you think? I can create a pull request if you think this change is worth implementing. Thanks! Gari