Setting up OMERO based virtual microscopy (...from Mirax)

Hi!

We plan to set up an OMERO based virtual microscopy platform, to be used in histology and histopathology teaching. We already have a basic OMERO server running.

What we need is simply means to easily publish & share slide scanner images on the web. Something like this: https://cloud.aiforia.com/ELTDK/

What do we need to add, and how?

This includes importing our current images into the OMERO system. Our images were generated using a 3DHistech scanner, so they are in the Mirax format. Our current virtual microscopy system converts them to the ECW format. Thus, these two formats are currently available.

Contacts to anyone who knows someone operating such a platform highly appreciated!

With best regards,
Mikael Niku
Veterinary biosciences
University of Helsinki, Finland

Dear Mikael

Thank you for your post.

Please see my answers in between your text below

What do we need to add, and how?

It is fully possible to establish an OMERO.server based on the one used in Dundee for teaching (also called “Virtual Microscope”),. The easiest way to replicate the whole server setup is to use the ansible administration software and the following ansible playbook . If you are not able or willing to use ansible, the ansible playbook is still useful, as it contains all the necessary configuration changes which are to be made on any OMERO.server to “turn it into” a teaching server with identical features to the “Virtual Microscope” in Dundee.

There might be a case (less ideal) that you want to add the features of “Virtual Microscope” to an existing OMERO.server, which would thus be a “mixed use” server (partially, perhaps, for research data, partially for teaching purposes). If this is the case, then please let us know, and can discuss the possibilities of such setup further.

This includes importing our current images into the OMERO system. Our images were generated using a 3DHistech scanner, so they are in the Mirax format. Our current virtual microscopy system converts them to the ECW format. Thus, these two formats are currently available.

There is no direct way to import your images in their original file format into OMERO. What you will need to do is to convert them into a supported multi-resolution image format e.g. OME-TIFF. A pipeline allowing to convert Mirax datasets into OME-TIFF is described at Converting Whole Slide Images to OME-TIFF: A New Workflow | Glencoe Software, Inc.

.The conversion is a two steps process:

  • First convert the images into “raw” images using the Command line tool bioformats2raw [1]
  • then run the Command line tool raw2ometiff to convert the “raw” images into OME-TIFF images.

The two Command lines tools can be installed directly from the GitHub repositories [1][2] or using conda [3][4].

Once converted into the OME-TIFF format, the images can be imported into your OMERO.server.

All the best

Petr

OME Team

[1] https://github.com/glencoesoftware/bioformats2raw

[2] GitHub - glencoesoftware/raw2ometiff: Raw format to OME-TIFF converter

[3] https://anaconda.org/ome/bioformats2raw

[4] https://anaconda.org/ome/raw2ometiff

2 Likes

I am too interested to set up OMERO for teaching purposes, on WSI plus some other images from radiology. I see the playbook mentions RedHat; while I can certainly go with RH, I have a number of Ubuntu servers around and I am more used to it. Do you think the playbook needs modifications to run, or can I consider to use it on Ubuntu as is? Thanks.

Hi Vincenzo,

yes, all our roles and playbooks are primarily tested and used on CentOS 7/RHEL 7 which is our operating system for most production deployments at this time.

As per the community feedback and contributions, there have been some efforts to make our roles and playbooks compatible with other Linux distributions especially Ubuntu/Debian. Which Ubuntu release are you using? For Ubuntu 18.04, there is some coverage e.g. the add-ons example playbook is tested both against CentOS 7 and Ubuntu 18.04 via Molecule. Ubuntu 20.04 is probably a bit more a work in progress but there are other people interested - see Install OMERO w/ ansible on Ubuntu 20.04

Hi Sébastien,
thanks for the info (and I’m on 18.04). However, regarding the above mentioned playbook, could CentOS be appropriate? In principle they should be equivalent.

Yes the playbook is tested both on CentOS 7 and Ubuntu 18.04 so you should be able to go for either platform.

At the end I went for an installation from scratch (single node), and I even succeeded :slight_smile: ,except for some small issue .
At present I have omero-gallery installed, but I was not able to use iviewer with it (just the basic viewer). I installed both, but I do not know if there is some extra configuration to set it also for omero-gallery.
A meta-question: i have heterogenous questions to ask about installation and preparation of slides. For the sake of future usefulness (of answers), what’s the policy of the forum: it is better to cumulate them under the same thread or have specific questions per thread? Thanks for all.

1 Like

Hi @VDM, congrats on the successul installation.

For the gallery/viewer interoperability, do you mean that when clicking on the eye icon, the legacy viewer opens? If so, it might be worth trying to set OMERO.iviewer set as the default viewer with

omero config set omero.web.viewer.view omero_iviewer.views.index

Re your meta-question, there is a fine-tuned balance and the answer might vary depending on the topic. We totally appreciate many questions in this forum arise from users trying to achieve a particular workflow and the chronology of a thread can reflect this. However, we find it easier to deal with separate threads for new question both in terms of tracking unanswered responses as well as improving the searchability for other users of this forum with similar question.

Very concretely, taking this thread as a good example, our discussion around the Ansible installation and platform support departed from the original question around using OMERO for pathology teaching and would probably have been worth a new dedicated post (and we could certainly split it a posteriori).

1 Like

Regarding the meta-question, here I will not add other questions, thanks.

I have already set this.
I am not sure if other settings interact with the iviewer. As example, if I login using the regular URL (http://XXX:XXX:XX:XXX/ ), this brings me to the default webclient, where iviewer has always been available. To reach the gallery, I have to login at http://XXX:XXX:XX:XXX/gallery/ , but there it is not.

Hi,
When you double-click to open an image in webclient (the URL will be `/webclient/img_detail/IMAGE_ID/) does this show you the iviewer?
If yes, then the same will be true in gallery.
If not, then the config is not working. Either it is set in the wrong environment or you need to restart the web server.

Sorry for the delay, I lost this message. No, I see /webgateway/img_detail/387/ . I restarted the web server, I saw a change in the home page, but now it does not load studies (however, I can go without issue to the “standard” site). Likely there is something wrong in the config; the log has errors inside, including:
2021-03-14 20:42:57,977 WARNI [ omeroweb.feedback.views] (proc.14457) handler404():200 Not Found: /about/js/version.js

2021-03-14 20:42:58,044 WARNI [ omeroweb.feedback.views] (proc.14464) handler404():200 Not Found: /mapr/api/config/

Ah, Apologies - I realise that you are using the ‘Default UI’ of omero-gallery, whereas I was thinking of the Categories UI since that has been the focus of more recent work and more people using it lately.

It will be a simple fix to update Gallery to use the webclient viewer for the Default UI.
I’ve created an issue at Use webclient viewer for Default UI · Issue #82 · ome/omero-gallery · GitHub

In the meantime, you might consider whether the Categories UI which is in use at http://idr.openmicroscopy.org/ looks good for you?

Regards,
Will.

Thanks Will, actually Ialready tried that too (at least I think :slight_smile: ).
In my config there is this line:

omero.web.gallery.category_queries={
  "all":{"label":"All cases", "index":0, "query":"FIRST50:Name"}
  }

So, with the config you have there, do you see the “Categories UI”?
For me, testing that same config on my local server, I see a this, (notice when I hover over the ‘eye’ icon, the link shown is /webclient/img_detail/201/.

Where do you see /webgateway/img_detail/387/?
Can you share a screenshot?

Thanks,
Will.

I think I messed up things. After having set the categories UI, it indeed shows the top part of your screenshot, but then there is only a line "Loading Studies… ".
Before that (and again now that I removed the omero.web.gallery.category_queries property, the interface was slightly different:

However, I can live without it: the most for me is to be able to show descriptions and the few available annotations.

OK, so you’re back to the “Default UI”. If this serves your needs then that’s great. But it’s really quite a minimal UI that only allows you to browse Group → Project → Dataset → Image hierarchy (doesn’t show any annotations) and open the Images in a viewer. It was written as a demo app and hasn’t been updated since then.

The Categories UI was developed for IDR to provide an improved entry-point to the webclient UI. You’ll notice that when you click on a study at http://idr.openmicroscopy.org/ you’re taken to the webclient UI where you can access all available data and annotations.
I’ve just realised that the “Loading Studies…” issue might be caused by an omero-web bug that is fixed at fix end of subtree childCount by will-moore · Pull Request #249 · ome/omero-web · GitHub. This will be included in the release later today. It’s due to you not having any Screens (High Content Screening data).
Or you could create an empty Screen using the toolbar button: Screenshot 2021-03-17 at 10.21.45

Hope that helps.
Will

I am slowly understanding the various capabilities :slight_smile: . In fact, I do not have Screens.
I will wait the next release and then check whether to use the Categories UI or the standard Webclient, thanks.

1 Like