Images from ROIs track xy

[will-moore]

From user request to create a Movie from an ROI when the ROI moves in x and y over time.

This PR improves the Images_from_ROIs.py script so that the X and Y of the rectangle at each time point are used to crop the image.

To test:

• Draw a Rectangle in Insight, then right-click and "propagate" it through Time (drag the square across Time axis to cover all the T-indices you wish to).

• Close that dialog, then update the X and Y position of the Rectangle at each time point (so as to track a moving object)
• Save the ROI (with multiple Rectangles within it). You can do this for multiple ROIs on the Images.
• Return to the main Insight window, select the Image and click on the Run-script button (green arrow with cogs) and find the Images_From_ROIs script (your hierarchy might look a bit different)

• You can run this script with the default values.

• A new Image should be created for (each) ROI and the crop at each time-point should respect the position of the Rectangle at that time-point. These will be in a new Dataset (called from_ROIs by default).

• NB: to export each of these Images as a Movie, use the Make Movie functionality for each one in turn.

cc @pwalczysko

[imagesc-bot]

This pull request has been mentioned on Image.sc Forum. There might be relevant details there:

https://forum.image.sc/t/creating-movies-from-rois-in-different-xy-positions-throughout-a-timecourse-in-omero/62776/2

[pwalczysko]

What is the next step if a movie is a desired end-product ? Combine images script ?

[pwalczysko]

Bug Taking the FRAP image in https://workshop.openmicroscopy.org/webclient/?show=image-28669 and not checking the Make Image stack checkbox, I still get the image stack.

Edit: Follow-up on the above: When I do check the Make Image stack checkbox, I do not get image stack, just one single plane.

Marking as a bug, it seems other way around than it should be, or rather, more precisely, in the case where I do not check the checkbox, I would expect many single images, which does not happen either.

[pwalczysko]

As discussed with @will-moore , maybe just adjust the description of the script.

the script principally does what we want, it is historically much more elaborate in comparison of with what is needed in a present case.

[pwalczysko]

Understood now what the script is doing. It is definitely catering for the new case and the behaviour is improved.

Ready to merge fmpov.

[will-moore]

Steps to update this on nightshade server:

• [ ] merge this PR
• [ ] tag this repo
• [ ] open PR to edit training playbook https://github.com/ome/prod-playbooks/blob/master/omero/training-server/playbook.yml to grab the tagged script and upload
• [ ] merge that PR, test deploy on a training server
• [ ] open a corresponding PR to add block to https://github.com/ome/prod-playbooks/blob/f124c1ec8e7562cf50a69b04576eb7e407b0ef1f/ome-dundeeomero.yml#L177-L184
• [ ] merge that PR
• [ ] decide on time-slot and send e-mail to notify nightshade users via web ui.
• [ ] run playbook to update

[will-moore]

To test the last commit - drag the Rectangle outside the bounds of the image. Script should run without an error.

[pwalczysko]

To test the last commit - drag the Rectangle outside the bounds of the image. Script should run without an error.

Works as described, it does not crash when ROIs are outside of the image in some timepoints. (could also repeat the crash with the previous script)

All seems good to me here.