Macro results vary over multiple executions

[haesleinhuepf]

Repeatedly calling the same macro code from pyimagej leads to different results.

See detailed description: https://forum.image.sc/t/macro-results-vary-when-calling-it-via-pyimagej/57744

[imagesc-bot]

This issue has been mentioned on Image.sc Forum. There might be relevant details there:

https://forum.image.sc/t/macro-results-vary-when-calling-it-via-pyimagej/57744/5

[ctrueden]

The Troubleshooting doc now mentions this issue as of imagej/pyimagej@e1b255450942dac1ea9a3e3f3afa85ee225c89b0. But no one has had time to solve it, unfortunately.

[imagejan]

I commented on the forum thread with some insights:

For what it’s worth, I was able to reproduce the issue within Fiji’s script editor (i.e., independent of #pyimagej). Running the following macro code repeatedly:

run("Blobs (25K)");
setAutoThreshold("Otsu");
setOption("BlackBackground", true);
run("Convert to Mask");
run("Analyze Particles...", "  show=[Count Masks]");
getStatistics(area, mean, min, max, std, histogram);
print(max);

… will print ever increasing values. It seems like there’s a counter for the Count Masks images that doesn’t get reset properly.

Judging from this line: https://github.com/imagej/ImageJ/blob/152c3d3def253a9c55f03214995b683f6f7008eb/ij/plugin/filter/ParticleAnalyzer.java#L1009-L1009

          case ROI_MASKS: drawRoiFilledParticle(drawIP, roi, mask, rt.size()); break;

… the value for each drawn ROI is determined from the size of the internal ResultsTable (rt.size()).

---

I can work around the issue by adding Table.reset("Results"); to my macro code:

run("Blobs (25K)");
setAutoThreshold("Otsu");
setOption("BlackBackground", true);
run("Convert to Mask");
run("Analyze Particles...", "  show=[Count Masks]");
getStatistics(area, mean, min, max, std, histogram);
Table.reset("Results");
print(max); // will now always print 64

[ctrueden]

@imagejan Would you consider this a bug in ImageJ? I.e. should we move this issue to imagej/ImageJ?

[imagejan]

Would you consider this a bug in ImageJ?

Yes. I just tested it with plain ImageJ, and observe the same behavior as seen with Fiji.

[ctrueden]

@rasband I transferred this issue here from PyImageJ. Based on your forum post I infer this might actually be working as intended, in which case this issue could be closed as not planned.

[rasband]

The particle analyzer is working as expected. The Count Masks are filled with a grayscale value corresponding to the particle number. To insure the particle numbers start at 1, use the “Clear results” option, as in this example:

  run("Blobs (25K)");
  setAutoThreshold("Otsu");
  setOption("BlackBackground", true);
  run("Analyze Particles...", "clear show=[Count Masks]");
  getStatistics(area, mean, min, max, std, histogram);
  print(max);