search

BICCN / cell-locator

manually align specimens to annotated 3D spaces
https://cell-locator.readthedocs.io
Other
19 stars 7 forks source link

User draw annotation on left, it appear on right in 3D #101

Closed jcfr closed 3 years ago

jcfr commented 4 years ago

Description

Reported by @wbwakeman

When users draw an annotation in the Slice viewer on the left, it appears on the right in 3D.

image

Expected behavior

Annotation should appear on the left

jcfr commented 4 years ago

Related to #74 (ReferenceView: Select sagittal -> 2D and 3D view have different orientation)

jcfr commented 4 years ago

Annotation now appears at the expected location:

image

deenick commented 4 years ago

@jcfr Thanks for making this change. The current view as currently stands means that users need to draw on right side of 2D image for a Left Hemisphere annotation on 3D image. Is it possible to flip 3D image so right is right and left is left? This would minimize errors/confusion I believe.

jcfr commented 4 years ago

Based on the following description of the atlas (available here), I looks like the right side is shown on the right.

Copied from issue https://github.com/BICCN/cell-locator/issues/48

wbwakeman commented 4 years ago

I think the point that Nick is trying to make is that the coronal views of the brain in the Cell Locator tool are shown as if we are looking at the front of the brain, like "face-to-face" with the mouse. If both views were rotated so that we were looking at the back of the brain, then when a tool user draws something in the "right" side of the picture, that is going in the "right" hemisphere of the brain (and vice versa).

I can imagine Nick's concern is that a user who has to think "I'm drawing a polygon in the right hemisphere, so therefore I need to draw it on the left side of this image" is going to make mistakes.

jcfr commented 4 years ago

After further analysis, the problem has been identified and fix is being along side new installer will be published shortly.

It is important to get this right so that we can finalize the LIMS integration

deenick commented 4 years ago

Thanks for providing a better explanation Wayne. I think resolving that will reduce errors when drawing.

jcfr commented 4 years ago

After further analysis, part of the fix for #48 implemented in 49425f929a148a was incorrect.

Commit 2910414 revisits the approach and ensures the right part of the viewer is the right part of the brain.

image

image

jcfr commented 4 years ago

@deenick The installer available here should address the problem, see https://github.com/BICCN/cell-locator/releases/tag/0.1.0-2020-04-30

deenick commented 4 years ago

@jcfr I just downloaded 4.30 version using the installer link above. I do not recognize the user interface with this new version, it appears completely different. Unable to figure out how to see slicer or draw polygons as done with previous versions.

jcfr commented 4 years ago

Could you share a screenshot ?

jcfr commented 4 years ago

Also you could right click somewhere near the top of the application, and show the python interactor. Is there any information displayed there ?

image

jcfr commented 4 years ago

Here is the issue:

image

Note all dependencies have been updated while building the project. I will address this and also add a testing step in the release process to prevent future regression like this one.

Please, wait to hear back from me.

deenick commented 4 years ago

When first opens: image

deenick commented 4 years ago

After selecting all options for module panel: image

jcfr commented 4 years ago

@wbwakeman @deenick The installer has been fixed. Consider re-downloading from https://github.com/BICCN/cell-locator/releases/tag/0.1.0-2020-04-30

deenick commented 4 years ago

Thanks @jcfr! I just downloaded and looks good at first pass. Both views are same so left is left. This small change will be very helpful for my team to annotate dissections

jcfr commented 4 years ago

Thanks for the update, but while working on documenting the meaning of raw/pitch/yaw, I realized that there is still something off.

If you rotate the volume to match the orientation of

Available at https://help.brain-map.org/display/mousebrain/API#API-InformaticsDataProcessing

where S is P and I is V, you can see that the the dataset is not properly oriented.

jcfr commented 4 years ago

I will have an other look at the header information in the nrrd file, as I discussed in the past with David, the info are incorrect .. see https://github.com/BICCN/cell-locator/issues/48#issue-386294514

wbwakeman commented 4 years ago

@jcfr I just tried this on the release from yesterday and it doesn't actually seem to work. Here is a screenshot of my drawing on the left side. image

deenick commented 4 years ago

@wbwakeman It works as expected for me image

wbwakeman commented 4 years ago

Thanks @deenick . I'll try again, double-checking that I am not doing something silly

wbwakeman commented 4 years ago

I think it was because I had started by loading one of your teams 10X annotations. Then I started a new drawing but the view was still reversed. So that is something to take note of. Although the location of the annotations that your team has already done are in the correct hemisphere, the view in the left pane still appears as reversed.

deenick commented 4 years ago

@wbwakeman Correct. I had my team load in old jsons and they said the view on right flips back and is reversed when viewing drawings made using previous version.

wbwakeman commented 3 years ago

I'm thinking this one can be closed now. What do you think @jcfr ?

allemangD commented 3 years ago

I agree. I'm not able to reproduce; this should have been fixed by #51 and #168. Integrating @lydiang's work in Fixing direction and spacing in CCF volume files.ipynb will supersede this anyway. Closing.