Conversion of elements from PALM Zeiss laser micro-dissection system to ImageJ domain

Hello everyone!

I have scanned the forum but could not find any relevant question. Has anyone successfully converted the coordinates from elements file which is exported via from PALM Zeiss system into ImageJ coordinates?

Background:

I am working on a project in which a user obtains two samples, one stained and one unstained using the Zeiss Laser Microdissection. We then draw a ROI on the stained sample (using Zeiss software) and export the coordinates in the form of an elements file. This is a text file containing all the coordinates of the ROI outline and some metadata. Externally in FIJI, we try to align the stained and unstained samples via registration. Using the obtained transformation, we transform the coordinates in the elements file from the stained sample. We write this to a new elements file and import it into the Zeiss system. This transformed ROI is then placed on the unstained sample. In the ideal case, it should be at the same location as in the stained sample.

After a lot of experiments and tests, I have managed to normalize the coordinates in the elements file to the stage coordinate system after following https://github.com/fabricecordelieres/IJ-Plugin_PALM_Zeiss_workflow/blob/master/Docs/Poster_Szeged-Hungary_NEUBIAS_2018.pdf but after that its still a mess. The ROI is clearly offset which could be corrected by hardcoding some value. But what I am struggling with is the scaling. Some images below illustrate the problem.

How the ROI looks like in Zeiss (Rectangular ROI on left):

How the same ROI shows up in FIJI after normalizing wrt stage coordinate system:

Any help is appreciated! Thank you!

Best,
Gayathri

1 Like

Hi Gayathri
I think the image calibration is wrong in the Fiji image you’re showing. The calibration info shown in the image is 238125x177800 microns which is 23x17 cm, which is wrong by about a 10 fold factor if you consider the pen stroke is about 1mm.
This might ruin the ROI coordinates.
Jerome.

Hey @jerome!

Yes, you are absolutely right and this was the main issue with the system that it did not give provide correct pixel width in the metadata. Or more so, it was in the metadata in an indirect way. The correct image size in microns and pixels was available in a section called “Description” when you open the OME-XML metadata while importing the image using bio-formats. By using this information, we were able to figure out the correct pixel width. We knew that we had to normalize the coordinates with respect to the stage coordinate system but found that it was inverted, i.e. the 0,0 was at the bottom right. Thanks to my colleague who figured this all out we now have a working pipeline to read the coordinates, transform them and export new coordinates.

Best,
Gayathri