Feature Question: Automate marker extraction from OME-TIFF metadata for mcquant

[PranavMotarwar]

Hi MCMICRO team,

While running MCMICRO on PhenoImager .qptiff data, we encountered a recurring issue where mcquant throws the following error:

Exception: The number of channels in markers.csv doesn't match the image

We were able to resolve this by manually extracting the channel names from the OME-XML metadata embedded in the .ome.tif file (e.g., EVA_TMA_Scan1.er--unmicst.ome.tif) using tifffile and writing them into a markers.csv file with the format:

marker_name
AQP1
Axl
CD11b
...

This step is a bit manual (especially with large batches). Is there any way to:

Include a utility step or an optional MCMICRO module that:

• Parses the OME-TIFF header metadata,
• Extracts channel names,
• Automatically generates markers.csv (or equivalent),
• Ensures alignment with the input image.

This would help:

• Avoid pipeline crashes due to mismatched or forgotten markers.
• Automate what is currently a side-script and prone to inconsistencies.

Let me know if something like this already exists or if you'd suggest a better solution.

Thanks so much for your help!

Best,
Pranav Motarwar
pm3891@nyu.edu

[ArtemSokolov]

Hi @PranavMotarwar,

I wonder if it would make sense to add your script to the mcquant module, where it would try to extract channel names from the image if a markers.csv is not provided. We can then make markers.csv optional in MCMICRO.

Do you have the script on GitHub somewhere, so I can take a look at it?

[PranavMotarwar]

Hi @ArtemSokolov,

Thanks for the reply. I agree that making markers.csv optional by auto-generating it from OME-TIFF metadata would make MCMICRO even more user-friendly.

Below is the Python script I used to extract channel names from the OME-XML metadata of .ome.tif files:

`from tifffile import TiffFile
import sys

if len(sys.argv) != 2:
print("Usage: python extract_channels.py <path_to_ome.tif>")
sys.exit(1)

ome_path = sys.argv[1]

with TiffFile(ome_path) as tif:
ome_metadata = tif.ome_metadata
xml = tif.series[0].metadata
channels = [c.Name for c in xml['Channels']]

print("marker_name")
for channel in channels:
print(channel)`

Best,
Pranav Motarwar
New York University

[ArtemSokolov]

Hi @PranavMotarwar,

I'm a little puzzled by your script. You're extracting ome_metadata, but then you're not using it. Instead, it seems you're reading channel names from the metadata of a TiffPage (series) object.

I just pushed a prototype that looks at the .name property of the Channel objects in OME metadata and falls back onto the .id property if .name is not initialized. I followed the ome_types Channel specification.

Can you check if this will extract your channel names? I'm a little worried that yours are not stored in the ome_types.Channel objects.

You can install the prototype directly from GitHub with:

pip install git+https://github.com/ArtemSokolov/quantification

and then run it with

mcquant --masks /path/to/mask.tif --image /path/to/image.ome.tif --output .

There's now a print statement that will display the channel names it extracts. If that doesn't find your channel names, we may have to add a special case for your images, but I worry that the logic won't generalize. Do you have control over where the metadata is written in your OME-TIFF files, or is that predetermined by the instrument?

If everything looks good, then I will package this into a new container version and update MCMICRO to use it.