Opening large multichannel *.czi pathology images

Hi,

I have a question regarding large multichannel “whole slide” pathology images. I am working with virtual slide images acquired on the Zeiss AxioScan system (*.czi files). There are 4-5 color channels in each image and (in one case) multiple z-planes.

As far as I understand it right now, in order to do analyses in Cellprofiler on these images, I have to export parts of these huge images as tiles and then import the tiles into Cellprofiler. I managed to do that. However, since I want to carry out further “downstream” data analyses with FCS Express, I need to rename the images according to a specific naming scheme (giving row and column numbers). Doing this manually is really not feasible with the number of images I have to analyze. The Zeiss Software (ZEN blue) seemingly does not offer the possbility to apply a custom naming scheme to the tiles, even though I can tell the software exactly how many rows and colums I want. Still, ZEN blue merely assigns continuous numbers to the tiles.

Is there any way to make Cellprofiler open the *.czi images directly, generate its own tiles and carry out the analysis in a way that is compatible with further analysis in FCS Express?

Any suggestions would be greatly appreciated!

Hi,

Yes, you can load .czi files directly. Take a look at the help here in CP: Help menu > Creating a Project > Loading Image Stacks and Movies.
The procedure is a little complicated, but I have done it for some czi files myself. If you are still having trouble, please post a sample file, as it is possible yours is encoded differently than what I have seen.

Cheers,
David

Hi David,

thank you for your reply. Sorry that I did not notice it for some time. I have now activated e-mail notification.

I had tried to load a *.czi file in the past but it did not work. Following your post I have now tried again with a different *.czi file and now indeed it worked nicely.

I can see the metadata and I can work with and analyze the individual tiles (sized 1600 x 1600 pixels) contained within the *.czi file. Even the channel names (e.g, DAPI) are correctly recognized.

However, I am trying to analyzed the data generated by CellProfiler with FCS Express 4. FCS Express has an import option for CellProfiler data but it needs to know the rows and columns of the individual image tiles. Unfortunately, this information does not seem to be read out by CellProfiler from the Zeiss metadata. The information must be stored somewhere within the *.czi file because I can open the file correctly “stitched” in other software, e.g., in Image Pro Premier.

So, my question is: is there any chance that I can do my analysis directly from the *.czi file AND still be able to analyze the data in FCS Express ?

The alternative would be to export large images from ZEN blue in tiff format, slice them into individual tiles using a tool named “total image slicer” (recommended by De Novo software), import and analyze these tiles in CellProfiler and then export the data to FCS express. It only seems so complicated to me, since the *.czi already contains the tiles.

Any ideas?

Kind regards,

Arnulf

Hi Arnulf,

I see what you are saying. As our Help says: “Since the metadata is often image-format specific, this [extract metadata from headers] option will extract information that is common to most image types…”. But I will talk to our software engineer to see if there is a way to expose the rest of the metadata to the user, including tile position.

In the meantime, would you mind posting a sample czi file from your data here, along with the expected tile sizes? If we can do it easily, this will help guide us.

Thanks,
David