stitched, fused czi opening as series

[ebbwdan]

Hi Melissa,

I'm sorry to trouble you again with this but I am still having issues with czi files. I have uploaded a file to your ftp server ('IP-Stitching-01_140716.czi') to show what I mean. I stitched and fused this image in Zen but still when I open with Bioformats it thinks that this is a series with a number of files equivalent to the number of tiles. I don't know whether there is anything in the meta-data which says it has been fused but it definitely has been so should open as a 2D plane (with 3 channels in this case).

We are just getting our Omero server up and this is proving to be a bit of a stumbling block for us since a large chunk of the data produced here is in this form.

Many thanks in advance for taking a look at this for me!

Cheers,

Dan.

[rleigh]

From what I can see in Zen and OMERO, this image is 9 tiles, so I don't think it's "fused", which I think should make it into a single large image plane. Would it be possible to double-check that?


Thanks,
Roger

[ebbwdan]

Hi Roger,

Yes, it appears as if the dataset is not fused, but I can assure you I chose this option in Zen. I did suspect there wouldn't be anything in the meta-data to flag that it had been fused. In addition, the 9 tiles that get imported to Omero (and indeed opened by Bioformats in ImageJ) are not displayed correctly. I guess this could be a Zen issue too so I will report also it to Zeiss.

Cheers,

Dan.

[rleigh]

Looking at the CZI file with "strings", I see

Code: Select all

<StitchingMode>StitchAndFuse</StitchingMode>

in

Code: Select all

<CustomAttributes>

. Looking at our CZI reader, we don't do any special handling of this element, so we can't at present handle the tiled data to render it as it would appear in Zen. I'll check with my version of Zen and see if there are any alternative ways to accomplish this.

Regards,
Roger

[ebbwdan]

Hi Roger,

Thanks again for taking a look. It turns out that the version of Zen I was using didn't have the latest hotfix (7); with this hotfix a stitched, fused image now opens correctly using Bioformats (I guess hotfix 7 relates to fusing the tiles).

However, despite opening using Bioformats through ImageJ, the image is refusing to display in Omero. It gets uploaded correctly and is processed with any problems but I get a message saying "not a valid image" when I try to open in the image viewer. I am uploading an example (should I start a different post about this?).

Cheers,

Dan.

[rleigh]

Hi Dan,

If it's working in Bio-Formats but not OMERO, I would suspect it's using a pixel type such as float which the server-side OMERO rendering engine does not yet support (we hope to address this in the future).

A new QA upload is fine to send us the file if you'd like us to take a look at it though. It would certainly be useful to see how Zen changed the file when fusing before and after the update.


Thanks,
Roger

[ebbwdan]

Hi Roger,

I have already uploaded a file using your ftp server (I have had problems with the QA upload failing due to the large size of the files). I was just asking if I should start a new forum post to specifically address the issue that stitched, fused mosaic images saved as CZI do not display in Omero.

We are very enthusiastic supporters of OME and think you guys are doing a brillaint job. So, just so you understand my urgency, we are a large institute (The Queensland Brain Institute) with a large variety of microscopy instrumentation (point scanning and spinning disk confocal, super resolution, slide scanning) and our IT manager has been working hard to convince the University of Queensland to invest heavily to allow us to implement Omero through a fully scalable cloud platform. We have several large research groups who are imaging brain tissue and are generating a lot of data of the form we have been discussing - large mosaic images (often times with multiple z-planes) saved as CZI. We were really hopeful of making our Omero server available to the whole institute in the next couple of weeks but I now I have a problem that a large proportion of data put onto the server will not be viewable.

I would appreciate if someone could test the file I uploaded to confirm that the rendering engine does not support the pixel data. Maybe a ticket could be opened so that I have an idea of a time-line for a fix? In order not to delay the roll-out of the server to the whole institute, I'll need to inform my users that they will need to use tif/ome-tif to upload data of this sort while the bug is being fixed.

Cheers,

Dan.

[rleigh]

Thanks for uploading the data. Sorry for the delay in replying, I have found and tested it now. The fused image from the newer Zen imports and renders just fine, being standard 16-bit data.

For the tiling issues, I have opened these tickets:
https://trac.openmicroscopy.org.uk/ome/ticket/12487
https://trac.openmicroscopy.org.uk/ome/ticket/12488
https://trac.openmicroscopy.org.uk/ome/ticket/12489
This will require some changes to our data model in order to be able to accurately represent tiles, and also to the rendering engine to be able to display them.

For your other questions, we'll get back to you shortly.


Thanks,
Roger Leigh