BUG - GE Deltavision Ultra Files

We have a GE DV Ultra (new scope they have) and it has plate reading functionalities. And it’s the images with plate reading that have a problem in bioformats.

  • I tried several versions of bioformats (with python-bioformats, pims, ImageJ, directly), including latest 6.1.1.
  • Java from ImageJ is jdk1.8.0_172

Issue: The microscope organizes the file with Channels, Z’s, and P’s. P is the visit point within a well. Acquire Ultra uses one file per well.

In the example in question, it’s a projection, therefore single Z.
Bioformats organizes with Channels (4 in example) and Z (25), the 25 are actually the 25 visit points within the well.
The problem is that the order and organization Between Points (Z in bioformats) and Colors is not correct.
I even tried to iterate ‘c’, ‘z’, ‘zc’, ‘cz’ in Python Pims and it’s in an incorrect order (easy to visualize the bad ordering with ImageJ).

Example file can be found in (it’s a projection to start simple):
https://northwestern.box.com/s/i82di7yyfl9ozb1iivwoc4hr9eqd1bwx

How it should look like
Point 1:
Point1|616x500
https://northwestern.box.com/s/3516kma26ntcbo19ne893nl4aki8rfcr
Point 3:


https://northwestern.box.com/s/3516kma26ntcbo19ne893nl4aki8rfcr
Point 7:

https://northwestern.box.com/s/kmflqai6lqqfw1g9qda2ys7cqv8fkck5!

How it looks like in ImageJ/Bioformats Plugin and with javabridge loci-tools in python-bioformats or PIMS

Z1


Z3

Z7

Thank you for any help. I need to analyze this data with my python scripts and my only solution now is to save every single frame in Tiffs with Softworx.

Thanks for the careful comparison. This does sound like the underlying reader in Bio-Formts 6.1.1 does not understand your data (as opposed to any issue with python-bioformats, pims, or ImageJ), but I’m hesitant to call it a bug. It sounds very much like either a change in the format (a “variant”) or perhaps a new file format entirely.

In either case, I’ve created issue 3394 to track any changes that may be made, but quoting from a similar thread:


That being said, those more well versed in the ImageJ API may be able to help you adapt your ImageJ workflow for working with this extra P dimension.

All the best,
~Josh

GE told me they paid someone to update this.

I managed to open the data from exported tiffs for now. I could even save them from python, To separate one-tiff files with 4 colors. However, I didn’t do a good job saving the Metadata.

Thanks for your reply.

A set of fixes for reading new Deltavision Ultra files has been proposed here:

That is not yet included in a release of Bio-Formats, but once it is the example file should open with the correct dimensions.

2 Likes

Thank you so much Melissa. If you want me to test it (I have many data sets) and can provide me a loci_tools, I’ll be more than happy to!