Importing colors

[ppouchin]

Hello everyone,

I currently work in Clermont-Ferrand, in France, as a computer engineer in a biology lab (GReD) and so, I recently decided to install your amazing OMERO software on our brand new server.

However, no one is using it yet here: they need some training, and I first have to understand everything and be prepared for their expectations.

I really like your interface and how everything is simple (except ICE installation maybe ). Nevertheless, there are still two things bothering me before letting the biologists use OMERO, and I can't seem to find the informations I need:

• Is it normal that when you import an image the custom colours defined during acquisition are erased ? Is there a way to prevent it ? I know how to change them in insight, but still...
  
• Apparently, the OME-TIFF files I export can't be opened in Imaris, nor in Volocity, and I think there's a problem even in ImageJ with bio-formats importer. Could I have messed something up during the installation ? The sample files from here do work (with Imaris, ImageJ and Volocity 4.2)...

If I can get my users to easily import and export what they're used to, I think OMERO will satisfy all their needs... (and they will stop walking around with their external hard drives...)

Thanks,
Pierre


Edit: I should have guessed sooner the problem with OME-TIFF files... I checked the schema version of the files from the wiki and it's not the one from June. And I checked my files, they are OK. I think I will put a script to export in an older OME schema, if it's possible, so people can use it with the different softwares (like old Volocity). This schema for example seems to work with Volocity 4.2... (It's the schema from the files on the wiki.)

[wmoore]

Hi Pierre,

We do not intentionally 'erase' colors on import - it's just that we have to work out how the file format is storing colors and manually set the colors in OMERO. It's different for every proprietary file format so we haven't done it for most of them (many of them don't store this info at all).

So far, we've taken a different approach of setting the initial colors based on the wavelength of acquisition.

If you could let us know the file formats that you are interested in, we could see if it's possible to read the colors on import.

Many thanks,

Will.

[ppouchin]

Thanks for your reply.

Here, they mostly use LIF and LSM files from Leica and Zeiss (albeit there might also be Olympus Multi-Tiff files occasionally).
But it's not that important : it would be interesting but I think they would understand if they had to copy/paste their settings onto whole datasets, which is not really time consuming.

Concerning the exports, I'm going through the docs I can find, but it's really vast... Is there an easy way to export to an old ome-tiff schema (for legacy purposes: even the latest Imaris can't open recent ome-tiff files, so our old licensed software won't do better...) ? I've seen the generateXml function but I suppose the schema is hardcoded somewhere... Would I be able to copy that function (that I don't find...) to a new script, but using a different schema instead ?

[wmoore]

Hi Pierre,

I'm afraid we don't support export to old schemas.
We use the Bio-Formats library to map between the latest OME data model and the OMERO database so it's pretty much hard coded. http://www.loci.wisc.edu/software/bio-formats

Do you currently use Imaris and Volocity to open your LSM & Liff files (or older OME-TIFFs)? One solution is to archive these original files when importing to OMERO, then you can always export these files for use with Imaris & Volocity - see a demo of this http://cvs.openmicroscopy.org.uk/snapsh ... ImageJ.mov

The only problem with this approach is that it doubles the storage space used in OMERO.
Sorry I can't be more help. We'll discuss support of old schemas but can't promise we'll change strategy any time soon.

Will.

[ppouchin]

I see. Well, I guess I should tell the users to archive their files, but using twice the space we'd use otherwise worries me a little. But the upside is that scientists would keep their original data that way...

Anyway, I understand that as a standard format there should be only one possible format to export to. However, considering how long it takes for private companies to accept a standard format, and adding the fact that labs can have old licensed softwares which are not always up to date, it could be nice to have the possibility to choose the version of OME when you export. Like OpenOffice with "swx" and "odt", or Excel with "xls" (5.0, 97-2003) and "xlsx" (2007)...

But I will still try to do something. Even though it doesn't look that easy...
I'm afraid that, apart from using bio-formats ImageWriter class, adding "generateTaggedTiff" and "do_taggedTiff" functions to ExporterI.java and recompiling, there's not much I can do...

Edit: I tried to change things in the source, but well... It's not a 5 minutes business...
So I thought I could use the LOCI command line tools to convert ome-tiff to an other format and tried it.
However, I discovered that the dimensions seem to be in the wrong place: whenever I convert a file with the tools or open it in ImageJ, slices and channels intertwines... where as this file works perfectly... I suspect a Big/Little Endian problem since I don't see other meaningful differences... But I still don't see how to solve it.

[wmoore]

Hi Pierre,

So, we discussed this issue today and decided that we should support export of ome-tiffs that are compatible with old schemas. This would probably not be all the metadata, since conversion from the latest model to old models is tricky or impossible and probably not necessary for most people's needs. If you simply need to open the file to work with the image data, then you don't need all the metadata.

We will probably try to do this using an xml transform to convert the essential data <Image><Pixels> etc from the latest schema into an older schema and replacing the XML tiff header with this transformed XML. However, we are unlikely to release this solution for a number of months.

I haven't tried your ome.xml sample file but I don't understand exactly what you are trying.
Are you exporting from OMERO and opening with ImageJ - using the OMERO 4.2 and the latest BioFormats plugin (or an older BF plugin)? How did you generate the sample file? It's possible that you are getting the dimensions mixed up by using different versions of BioFormats in OMERO vv ImageJ?

OK - so I see it's our sample file! I just tried to import it to latest OMERO and you're right - All dimensions become T. Created a ticket for this http://trac.openmicroscopy.org.uk/omero/ticket/3668 .

[ppouchin]

So one day, we might be able to export images and basic metadata into an OME-TIFF file compatible with old software ? That's really great ! Thank you.

About the file I imported, yes it's your sample file. When I import it in OMERO 4.2.2, everything seems correct in Insight.
The file I attached is what I get when I export your sample file afterwards.

The original opens correctly with ImageJ (latest version of BF I think), the imported-then-exported one mixes up Z & C, but I didn't see why...

[wmoore]

OK - I've reproduced that error and updated the ticket above. Please cc yourself (e-mail) on the ticket if you want to be notified of updates to the ticket. Thanks,

Will.

[ppouchin]

I'd do that... But even connected as "omero" all I can do is post a comment... Should I post a comment with my email ? Sorry, I'm not used to trac...

Edit: Thanks

[cxallan]

Not a problem. Will has added you to the CC line as you have no doubt now noticed. This should let you know when we've made any progress.

Thanks for your feedback!