Longest Shortest Path Outputs Inconsistent

[ali2019fau]

Hello,

Describe the Bug

The longest shortest path outputs are different across machines. Mainly, there are a different number of rows in each table.

I tested two Windows and one Linux computer running ImageJ 1.53j / Java 1.8.0_172 and they had identical results. Then, I tested two MacOS’s running ImageJ 1.53j / Java 1.8.0_202 and they had different results from the three above and from each other.

Expected Behavior

The outputs should be identical.

Steps to Reproduce

1. Download and open this image: https://drive.google.com/file/d/1p6jmT2hmJWZToSDhaOGl8-oAX904zqpq/view?usp=sharing
2. Open Interactive Interpreter and copy and paste the macro (ijm form) below

   dir = File.directory;
   run("3D Objects Counter", "threshold=50 slice=5 min.=5 max.=134560 objects centroids statistics summary");
   selectWindow("Objects map of Substack (36-45).tif");
   run("Skeletonize (2D/3D)");
   run("Analyze Skeleton (2D/3D)", "prune=[lowest intensity branch] calculate show display original_image=[Substack (36-45).tif]");
   selectWindow("Results");
   saveAs("Results", dir + "Results" + ".csv");

3. Run the macro. Output should be where the image is.

Environment & Results

Windows/Linux Results: Results (windows-1.53j, 1.8.0_172).xlsx

• ImageJ ver. 1.53j
• Java 1.8.0_172
• Mean of Longest Shortest Path Excluding Zeros: 7.284857

One of the MacOS Results: Results (macOS-1.53j, 1.8.0_202).xlsx

• ImageJ ver. 1.53j
• Java 1.8.0_202
• Mean of Longest Shortest Path Column Excluding Zeros: 3.380945

Thank you, Ali

[imagejan]

@ali2019fau did you check whether the global settings on both machines are the same?

What about the state of the Black background option in Process > Binary > Options...?

[ali2019fau]

@imagejan What do you mean by global settings? Is it the global settings within the program? What exact settings do you think would affect results like this? For now, I just checked over the IJ_pref.txt file for Windows and Mac and I am not sure if there is anything conclusive about it. I posted it here in case anyone wants to look it over too.

IJ_Prefs_Windows.txt IJ_Prefs_Mac2.txt

The Black background option is unchecked for both machines. Checking it and unchecking it does not change the results as far as I can see.

[imagejan]

Can you share your result image "Object map of Substack (36-45).tif"? When I run the run("3D Objects Counter") line of your macro, I get an image with the objects and their labels (numbers). I don't know if that is what you want to skeletonize.

If the issue is specific to Analyze Skeleton, can you share the skeletonized image you're working on, and make sure it's the same on both systems you're trying?