Multi-scene CZI files

[dsudar]

Hi all,

With Bio-Formats' capability to read many new variants of Zeiss's CZI format, we just bumped into one variant that it appears to have trouble with: multi-scene files from the Axioscan. When trying to load one of those into Fiji, it says "analyzing" for a while and then nothing happens with a silent fail.

When trying to import such a file into OMERO 5.3.0 using the command line import, it runs for hours at 100% on one core but it never appears to complete and I can find nothing in the logs that indicate what is happening. If I split the file into separate scenes using ZEN, the individual scene images load into either Fiji or OMERO easily and quickly so the distinguishing thing appears to be the multi vs. single scene (and of course the size).

The smallest multi-scene file I have access to is over 2GB. Could I upload one of those to your server?

Thanks,
- Damir

[dgault]

Hi Damir,

Thanks for reporting, feel free to upload the sample file to our FTP and we will investigate further.

David Gault

[dsudar]

Hi David,

I've been able to figure out a bit more about the cause. It's not due to the multi-scenes per se, it appears to be related to multi-scene files created with a more recent version of Zeiss's ZEN Blue for Axioscan. I'm uploading (QA record: 17711) a few examples: an older file "2015_08_21__16_09__0954.czi" that imports without trouble (and I verified that a much larger file that was created at that same time also works fine so it's not related to file size), and 2 files that were scanned today and which show the problem. When importing those with the command line importer the import process stalls with:

Code: Select all

dsudar@lincs2:~$ nohup omero import 2017_03_23__11_17__0017.czi &
dsudar@lincs2:~$ cat nohup.out
Using session edb4eae8-c071-41b4-96d3-2e33a3d3464c (sudard@localhost:4064). Idle timeout: 10 min. Current group: TestGroup
2017-05-03 17:25:00,322 147        [      main] INFO          ome.formats.importer.ImportConfig - OMERO Version: 5.3.0-ice36-b59
2017-05-03 17:25:00,330 155        [      main] INFO          ome.formats.importer.ImportConfig - Bioformats version: 5.4.0 revision: bdb4e1c50961e37431009f419b4c6b4ed8d17594 date: 20 March 2017
2017-05-03 17:25:00,355 180        [      main] INFO   formats.importer.cli.CommandLineImporter - Log levels -- Bio-Formats: ERROR OMERO.importer: INFO
2017-05-03 17:25:00,561 386        [      main] INFO      ome.formats.importer.ImportCandidates - Depth: 4 Metadata Level: MINIMUM


and while stalled like that, it uses 100% of a core, while doing nothing ...

Cheers,
- Damir

[mlinkert]

Hi Damir,

Thank you for uploading files, and providing additional details about the acquisition. I could duplicate the problem with both of the files acquired today.

We have received one other report of the same problem:

https://trac.openmicroscopy.org/ome/ticket/13320
https://trello.com/c/S60YCf0H/95-fiji-z ... ls-to-open

As far as we know, this only happens for a certain subset of .czi files with multiple scenes where each scene is large enough to require its own pyramid of lower-resolution images. Both the ticket and Trello card above will be updated once we have a fix; if you would like to be CC'd on the ticket just let us know.

Regards,
-Melissa

[dsudar]

Hi Melissa,
Thanks for the update. Yes, please cc me on that ticket. Let me know if you need any additional detail or examples. See you in Dundee.
Cheers,
- Damir

[blota]

Hi all,

Fiji seems to be able to open multiscene czi files from Axioscan. Thanks! I have two questions regarding these kind of files:

- How do I find which series are the top of the pyramid (in a macro)?
- All the "stages" of a pyramid have the same physical pixel size in their metadata. Is that a bug?

In case it's not clear, let me give an example. I have 8 scenes, but each has 4-6 subsampled version, so,when I load the file in Fiji, I have 50 series. I want to extract one channel of all scenes but I can't find which series correspond to my original full resolution scenes. I tried to look at the metadata, but the physical size x and y are not updated (hence the resolution is correct only for the biggest picture of the pyramid). I tried to look in the xml metadata but couldn't find any obvious way to distinguish two series inside the same pyramid from two series in different scenes.
(I'm Using Fiji with bio-format 5.5.2-20170529.105743-2)

Thanks

Antonin

[sbesson]

Hi Antonin,

- How do I find which series are the top of the pyramid (in a macro)?

Being able to group the various resolutions of a pyramid is the goal of the Bio-Formats subresolution API. If it is set using IFormatReader.setFlattenedResolutions(false), each series will correspond to a pyramid with individual resolutions being accessible using IFormatReader.setResolution(int).

In this case of the ImageJ/Fiji macros, this API is neither exposed in the macros - see https://trello.com/c/xWnT2XZC/67-imagej ... esolutions for the discussion. The only workaround at the moment might be to sort series by pixel sizes and find the largest one.

- All the "stages" of a pyramid have the same physical pixel size in their metadata. Is that a bug?

Absolutely, the expectation is that the physical pixel size should be increasing when moving towards lower resolutions of the pyramid. It is fair tI have captured this Zeiss CZI bug in Trello card and added to our backlog board.

Best,
Sebastien

[MSomerville]

Hi there,

We have a Zeiss Axioscan Z1 slide scanner and we generate multi-scene CZI files of whole slide scans. These CZI files operate as pyramids and are compressed via the Zeiss Jpeg-XR active lossy compression algorithm.

I have been trying to open these files using FIJI for a while. As it is FIJI rather than Image J I know my Bioformat plugin is up-to-date. So far I've had no success opening these files. We use Qupath software which also utilises the Bioformats plugin to open these files and this works fine.

I'm happy to send one of these files if someone wouldn't mind providing us with some help to generate a solution. Maybe there is a simple fix out there already?

Many thanks in advance for any help.

Michelle

[dgault]

Hi Michelle,

Do you have the stack trace or error message that is displayed when trying to open the files?

If you have a small sample file (<2GB) then you can upload it to https://www.openmicroscopy.org/qa2/qa/upload/, if it is larger then we can arrange for ftp details for you to transfer the file.

[MSomerville]

Hi there,

Sorry it's a silent error so I don't have any messages or info like that to pass on.

I have an image which I'll upload.

Many thanks for your help.

Michelle