I've used the "iterate" functionality on two tools now, the Import T1 MRI and the Morphologist 2015 (non-UI) pipeline itself. They work great! These will be great for running 100+ scans automated.
However, it would be nice if there was a way to set options globally for the whole set of iterations. For example, if I want to use NIFTI file format for the import on all, or if I want to do Sulci recognition on all imported subjects. It would be great to set global option once, rather than click on each iteration, set the option, click on the next, set the option, etc ...
Suggestion for "Iterate" functionality
- riviere
- Site Admin
- Posts: 1361
- Joined: Tue Jan 06, 2004 12:21 pm
- Location: CEA NeuroSpin, Saint Aubin, France
- Contact:
Re: Suggestion for "Iterate" functionality
Hi,
Actually to iterate a process, you can either use the "iterate" option instead of opening the process user interface, or open a process (in non-iterative mode), select some parameters, then chick the "iterate" button from it: all parameters set before iterating will be set the same on all iterations. This allows to set some parameters, but does not allow to set attributes leading (indirectly) to parameters (like format), however.
Moreover, if you need to iterate over hundreds of subjects, the "regular" Morphologist pipeline will become very slow during setup (at least) because it has many many parameters and we rely on an inefficient databasing system for parameters completion. There is a "simplified Morphologist" version (in morphologist/segmentation pipeline) which is the same pipeline but with fewer options, and in the shape of a single process instead of a complex pipeline. This one is thus more lightweight and can scale somewhat better for multiple iterations.
Morphologist-UI is based on the "future" pipelining system (named Capsul) and should not suffer such slowdown when iterating, but has the overhead of "direct" graphical display, so you probably lose what you gain on the other hand, and as you have seen this version is "new" and still has a few problems that we are still fixing. A non-graphical version of it ("morphologist capsul iteration") is also present but hidden (only visible in expert mode) because still more or less in beta-test mode. This one will be pushed forward in a next release.
Denis
Actually to iterate a process, you can either use the "iterate" option instead of opening the process user interface, or open a process (in non-iterative mode), select some parameters, then chick the "iterate" button from it: all parameters set before iterating will be set the same on all iterations. This allows to set some parameters, but does not allow to set attributes leading (indirectly) to parameters (like format), however.
Moreover, if you need to iterate over hundreds of subjects, the "regular" Morphologist pipeline will become very slow during setup (at least) because it has many many parameters and we rely on an inefficient databasing system for parameters completion. There is a "simplified Morphologist" version (in morphologist/segmentation pipeline) which is the same pipeline but with fewer options, and in the shape of a single process instead of a complex pipeline. This one is thus more lightweight and can scale somewhat better for multiple iterations.
Morphologist-UI is based on the "future" pipelining system (named Capsul) and should not suffer such slowdown when iterating, but has the overhead of "direct" graphical display, so you probably lose what you gain on the other hand, and as you have seen this version is "new" and still has a few problems that we are still fixing. A non-graphical version of it ("morphologist capsul iteration") is also present but hidden (only visible in expert mode) because still more or less in beta-test mode. This one will be pushed forward in a next release.
Denis
Re: Suggestion for "Iterate" functionality
Thanks for the reply Riviere.
I think I understand now. I was able to select the [...] button in the Morphologist iterate menu for sulci recognition and add "true", which then automatically selected sulci recognition for each subject.
But for Import T1 MRI iterate menu, I could never get Data Type or File Format unless I individually adjusted for each scan. Which I think you explained because it is an attribute instead of parameter. I will just have to live with it!
If auto-selecting for attributes is included in the Capsul Iteration for future release, it would be great.
Thanks for all the help!
I think I understand now. I was able to select the [...] button in the Morphologist iterate menu for sulci recognition and add "true", which then automatically selected sulci recognition for each subject.
But for Import T1 MRI iterate menu, I could never get Data Type or File Format unless I individually adjusted for each scan. Which I think you explained because it is an attribute instead of parameter. I will just have to live with it!
If auto-selecting for attributes is included in the Capsul Iteration for future release, it would be great.
Thanks for all the help!