Drop markups separately on the fixed and moving volumes

[lassoan]

When two volumes are very misaligned then setting the position of the dropped fiducial on the other image (where it was placed automatically) is almost impossible.

There should be an option to drop markups separately on the fixed and moving volumes.

[pieper]

This would be a big change to the logic and the workflow that I'd rather
not make.

I think it might be better to add a way to initialize the registration,
like a button for transforming the centroid of moving to the centroid of
fixed.

On Tue, Dec 17, 2013 at 3:21 PM, Andras Lasso notifications@git.luolix.topwrote:

When two volumes are very misaligned then setting the position of the
dropped fiducial on the other image (where it was placed automatically) is
almost impossible.

There should be an option to drop markups separately on the fixed and
moving volumes.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/3
.

The information in this e-mail is intended only for the person to whom it
is
addressed. If you believe this e-mail was sent to you in error and the
e-mail
contains patient information, please contact the Partners Compliance
HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in
error
but does not contain patient information, please contact the sender and
properly
dispose of the e-mail.

[lassoan]

Centroid may help sometimes, but not when there is a patient orientation mismatch.

No need to change the workflow. Just add a "Replace landmark" button (and "Delete landmark" and "Rename landmark" buttons, but those are separate tickets). If replace landmark is chosen then the next click in an image replaces the current landmark in the clicked image.

[lassoan]

Lack of separate placement/replace is a serious limitation. So far none of the 3 users that I tried to teach to use this could use it on the actual input images because of this. They can only use this module as a refinement step after they do the registration with Fiducial registration module or IGT / Fiducial registration wizard module.

[pieper]

Pull requests are welcome :)

Also we are having a project to go over ideas and issues so we can see how
things look.

http://www.na-mic.org/Wiki/index.php?title=2014_Winter_Project_Week:Steered_Registration

Note that Marcel is working on and has an example of direct manipulation of
the transform itself (not indirect as with the landmarks) that might be
more appropriate for this kind for initial 'ballpark' alignment.

On Tue, Dec 17, 2013 at 5:53 PM, Andras Lasso notifications@git.luolix.topwrote:

Lack of separate placement/replace is a serious limitation. So far none of
the 3 users that I tried to teach to use this could use it on the actual
input images because of this. They can only use this module as a refinement
step after they do the registration with Fiducial registration module or
IGT / Fiducial registration wizard module.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/3#issuecomment-30800082
.

The information in this e-mail is intended only for the person to whom it
is
addressed. If you believe this e-mail was sent to you in error and the
e-mail
contains patient information, please contact the Partners Compliance
HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in
error
but does not contain patient information, please contact the sender and
properly
dispose of the e-mail.

[lassoan]

Sorry, in this project I can only "help" with feature requests/bug reports for now.

I'm not sure transform manipulation could be a general solution, as it's very difficult visualize an unregistered CT and an MRI at the same time. Alignment by translation and rotation is also much more difficult for experts than picking 3-4 matching anatomical landmarks. But of course whatever will be implemented I'll try to use.

[lassoan]

I've tried to teach somebody today to use this module but it was a failure again. The issue was the same as reported above.

We wanted to register two MRIs of the same animal phantom before and after treatment. We located the same landmarks on the images but as soon as we clicked on the landmark point on the fixed image, the moving image slices were turned to black (because the selected landmark points was outside the moving image). Once the landmark point is dropped in a black area there is no way out: you cannot delete the landmark and as everything is black you have no idea where to drag-and-drop the fiducial.

The module is awesome, but I simply cannot teach any regular people to use it because the initial alignment problem is so difficult. If you could simply add the option to drop points separately on the fixed and moving image then the module would become usable.

[pieper]

Okay I hear you :)

How just having an initialization step like aligning the centers of the volume. This would be a lot easier to do (and probably also useful in its own right). The bookkeeping for the matched landmarks is a bit tricky and I think that adding the fiducials in two steps would be extra complexity.

[lassoan]

Initialization based on geometry or center of gravity would certainly help in many cases, so if it's much easier to implement it then please do it. Thanks!

[pieper]

The new behavior is that fiducials are clamped to the volume and are mapped through the current transform on placement. I find that this handles the reported issue to my satisfaction, although other methods are possible.