Open Microscopy Environment

Hello,

We're looking to create an open access atlas of mouse brain sections, with 4 immunofluorescence channels per image. We want to create a custom 'skin' for the web viewer to enable visitors to move between sections. We are employing a web developer to helps us achieve this.

Should we be looking at customising OMERO.iviewer (the histogram, ROI and brightness adjustments feature would be useful, e.g. we might want to have boundaries which we would otherwise be putting in an extra channel to enable these to be turned on and off); or should we stick with the standard viewer, and customise that?

If we want to use the Z slider for changing sections, do we have to create a huge image (we're using ImageJ for final prep before import to OMERO) with all sections in one file? One 16-bit 4 channel mouse brain section is 1.2 GB, and we might need 20 or more sections to cover our structure. How else would we get the viewer to navigate between images in an intuitive way, other than using Z? How easy would it be to have a film strip of thumbnails, or even a image of the brain from the side showing the rostral-caudal position of the current section with arrows to move in either direction?

Can anyone point us to open/public access uses of OMERO for atlas-type projects?

Thanks!

Hi,

Thanks for your interest.

In general I would encourage you to work with OMERO.iviewer since we have moved all our focus to that viewer and it has much better support for large tiled images. We are not working on the older viewer any more.

I'm not very clear on the customisations you want to make:

"A custom skin": Does this simply mean a different colour header and logo, or are there functional / layout changes you want to make?

"Move between sections": Scrolling through a Z-stack is supported, but yes, OMERO only supports this for a single image so you'd have to create a 'huge' image with all sections in one file. This should be doable by creating multiple OME-TIFF files and linking them via an OME-XML companion file. We can discuss the details in another post, but how are you currently preparing these files in ImageJ? Do you have a Macro/script you can share?

If you have each section as a different image within a Dataset (folder) then iviewer will display all those thumbnails and allow you to select them.

"boundaries": do you mean ROIs (polygons) or masks?

Orthoganal views of Z-stacks are possible in theory, but we'd need a fair bit more work on the OMERO server to support these, especially for big tiled images.

Customising the iviewer to do most of what you want shouldn't be too hard. The tricky part will be keeping your changes isolated from existing code so that its possible to port your changes to the next version of iviewer.

Unfortunately the principle developer of the iviewer has left the OME team, but I expect that the components of the viewer (built with https://aurelia.io/) can be reused in alternative layouts etc.

A mock-up screen-shot can help a lot in these discussion. We use github issues for this purpose (e.g. https://github.com/openmicroscopy/design/issues/91) so feel free to start an issue there if that makes it easier.

Regards,

Will

bmicklem wrote:Hello,
Can anyone point us to open/public access uses of OMERO for atlas-type projects?

I'm not quite sure what defines an "atlas-type project", but are you familiar with the Image Data Resource (using OMERO for public access images), e.g. https://idr.openmicroscopy.org/webclien ... ataset-369

Thank for your replies Will.

By "atlas-type", I meant something with aligned images across all sections, and a series of sections going all the way through a structure at frequent regular intervals; with an interface for moving between the sections. E.g. http://mouse.brain-map.org/gene/show/12432

Boundaries, if we were using iviewer, would be ROIs; if the standard web viewer would be an extra channels with white strokes around regions. Obviously, the ROIs would be much more useful in the future (e.g. analysing intensities in the regions, or realigning them with other equivalent position sections without the immunolabelling that defines the boundaries).

By 'skin', I mean branding, colour scheme, font, but also hiding unnecessary features (e.g. T slider), and navigation links back to other parts of our web site.

The navigation via an orthogonal viewer would not be calculated from the images, but would just be a (in our case) side view of the mouse brain (static reference image for all atlas datasets), with section positions marked on as grey lines, and the current position as a black line, and arrows to move back and forth. This would have to read in Z values from the images (these would be standardised stereotaxic coordinates), and draw lines in the appropriate positions.

When I have a chance, I'll mock up how we want our viewer to look. Thanks again, Ben

So we now have iViewer on our OMERO server, and the speed of zooming is much improved, and everything is better than the default viewer. The thumbnail viewer does indeed work nicely with thumbnails of other images in the dataset, and this should be OK for our atlas, in lieu of the custom side view of brain. Ideally, the thumbnail strip would have filenames, so we could label them by their distance from Bregma, but the hover to bring it up is not too bad.

Having all images in one giant stack won't work very well, as there are different reactions for different sections (e.g. brain is cut into sections and the series is collected sequentially into 6 vials, and vial 1 is chosen for a reaction with 4 primary antibodies, and vial 4 is chosen for a reaction with a different set of 4 antibodies); I did think of adding all channels to the giant file (8 for the previous example) and having black images for the missing channels in the sections which had the other set of reactions; but if we add many different combinations of antibodies in a series, it will get silly with the number of channels needed.

We've been having issues with the thumbnail browser not always appearing, which I think is due to the dataset not being owned by the public user (i.e it is there when logged in as the owner of the dataset, but not their if using anonymous public user (despite all images in the dataset being accessible directly by the anon user). I'll check that we can browse the dataset folder when we change ownership of the dataset. The main reason we want to browse the dataset folder in the file viewer is to see all metadata: we are using key/value pair metadata to list all immunohistochemical processing steps.

One thing that would be very beneficial is if these key/value pairs could be shown on the iviewer display.

OK, so I've created a new dataset owned by the public user, and still I cannot see the thumbnails in iViewer (all images in folder are owned by another user than the public one, but public has permission to see them).

See logged-in and logged-out PNGs attached.

Any tips on getting the thumbnails to appear in iViewer for the public user?

Hi Ben,

I think this is due to the public "omero.web.public.url_filter" regex. You've presumably set this to allow webclient and iviewer URLs, but the iviewer also needs to load images in the Dataset from "api" URLs e.g.

/api/v0/m/datasets/4/images/

You can see this by using the dev-tools in your browser (right-click anywhere > Inspect Element. Then look at the Network tab and refresh the page. This should show if there are any URLs that are getting rejected in red (filter by XHR to show only AJAX requests).

We'll need to improve our docs on this (they've not been updated since we've released iviewer).

When I tried to only allow webclient and iviewer URLs, I don't see thumbnails loaded in iviewer, but I do see the empty thumbnail panel (doesn't exactly match your screenshot), so there may be other URLs that are rejected in your case.

This is what's working for me:

Code: Select all

$ bin/omero config set omero.web.public.url_filter '^/(webadmin/myphoto/|webclient/(?!(script_ui|ome_tiff|figure_script))|webgateway/(?!(archived_files|download_as))|iviewer|api)'


Hope that helps.

Will.