Open Microscopy Environment

Hi All,

We are just getting started with OMERO and i have a few questions for you all.

1. What happens to the raw data? When i import images from various platforms into OMERO my understanding is that they are all converted to OME-TIFF files. If this is true then we need to also keep a copy of our orignial raw data, thus doubling our data burden right away.

2. Can images stored in the OME database be referenced directly via third party software? I.e is there a specific address/network location etc for each file?

3. Opening images - If i view RGB images in the built in viewer they come out fine. If i say view in IMageJ they open as a stack with each channel on a different plane. I can convert these back to composiute RGB but that is a bit annoying to do all the time.

4. Analysis - We do a lot of analysis with ImageJ/Fiji/MetaMoprh/MatLab/CellProfiler/Imaris/Volocity etc. Is it possibel to link into images with these platforms?

5. Scripting - The only information i can find about scripting in OMERO is about running python scripts on the server. Is this the only option? Can we write scripts in imagej etc and have them run on the serve too?


Thanks in advance

Cam

Cammyron wrote:1. What happens to the raw data? When i import images from various platforms into OMERO my understanding is that they are all converted to OME-TIFF files. If this is true then we need to also keep a copy of our orignial raw data, thus doubling our data burden right away.

You are roughly correct. But images are not converted to an OME-TIFF file: the metadata is placed into the database and the pixel data is placed in a binary file. But the data volume is doubled. We've been working to address this issue for some time, and the next release of OMERO will address this by only storing the original file on the server, reading the pixel data from this file.

Cammyron wrote:2. Can images stored in the OME database be referenced directly via third party software? I.e is there a specific address/network location etc for each file?

Each image has a unique ID on the server, so an image can be identified by this, along with the server url. Third party software can connect to the specified server url, and retrieve the image by ID. The webclient also provides a url to each image on the server: see viewtopic.php?f=5&t=2772

Cammyron wrote:3. Opening images - If i view RGB images in the built in viewer they come out fine. If i say view in IMageJ they open as a stack with each channel on a different plane. I can convert these back to composiute RGB but that is a bit annoying to do all the time.

The RGB planes are stored separately in OMERO, merged together in our viewers. I don't know much about ImageJ viewers or if there's a nice way to do this automatically.

Cammyron wrote:4. Analysis - We do a lot of analysis with ImageJ/Fiji/MetaMoprh/MatLab/CellProfiler/Imaris/Volocity etc. Is it possibel to link into images with these platforms?

The OMERO API is available to Java, Matlab, Python & C++, so this should support most of what you need. However, where you're relying on commercial software packages to implement the interaction with OMERO, you don't have so much control over this.

Cammyron wrote:5. Scripting - The only information i can find about scripting in OMERO is about running python scripts on the server. Is this the only option? Can we write scripts in imagej etc and have them run on the serve too?

It's possible to use Python as the 'glue' to other languages / packages. The easiest option for running ImageJ is to call ImageJ macros etc via the command line from Python.

Thanks for the quick response. I have had a look through some more info on the API etc. But one thing I am still struggling with is addressing the images in the database. I understand that each image has its own ID number but can you reference other parts like its original name or a sub string of it?

For example we could have a folder with say 20,000 images in it from a scan of a 384 well plate. The name of each image will contain the date of capture, the well number, the field number and the fluorescent channel number. Is it possible to address these in some way without having to bounce back and forward between omero, python, imagej and third party apps?

While it is possible to call imagej via python from the command line this is not the most flexible thing to do as.each imagej macro has to be hard coded for the specific analysis and.file locations.

I guess the stuff we are trying to be able to do has already been addressed by the commercial Columbus implementation of omero?

Cammyron wrote:I understand that each image has its own ID number but can you reference other parts like its original name or a sub string of it?

For example we could have a folder with say 20,000 images in it from a scan of a 384 well plate. The name of each image will contain the date of capture, the well number, the field number and the fluorescent channel number. Is it possible to address these in some way without having to bounce back and forward between omero, python, imagej and third party apps?

I'm not sure exactly what you're asking, but if you retrieve an image by ID, you can certainly get it's name and any other metadata associated with it. You can also query OMERO for images with certain names (and filter by Screen / Plate etc if needed).

Can you give a simple example of a workflow?

Cammyron wrote:While it is possible to call imagej via python from the command line this is not the most flexible thing to do as.each imagej macro has to be hard coded for the specific analysis and.file locations.

I need to make some better examples of Python / ImageJ interaction, but there's really no restriction since Python can write the Macro to disk and pass it to ImageJ, along with any command line arguments. You can get a flavour of this in a demo movie https://trac.openmicroscopy.org.uk/ome/ ... ection.mov (ImageJ is being called from the Python web code, not the scripting service, but the principle is the same).

Cammyron wrote:I guess the stuff we are trying to be able to do has already been addressed by the commercial Columbus implementation of omero?

They certainly have some nice analysis workflows on to of OMERO, but if you have existing analysis that's working for you, it would be nice to hook it up to OMERO.