Scaling for analysis of average size (.czi to .tiff)

Hi,

I am doubting if the scaling that I use for measuring areas is indeed correct and if it is the best way to do.
I first drew a line from the left to the right corner in my .czi file and measured the distance in µm with the zen blue software.
To write the macro, I opened one of the tiff files (czi converted to tiff) drew the same line and set the µm distance that I got from the czi image.
Each image consists of several tiles (4x4) stitched together taken with a 5x objective.
Is this a good way to do the scaling? If so, can I check somewhere in the summary table if indeed areas are expressed in µm2 rather than pixels or inches?

This is my macro:

run("Set Scale...", "distance=5000.4052 known=6455.941 unit=inch");
run("8-bit");
setAutoThreshold("Default dark no-reset");
//run("Threshold...");
setThreshold(11, 255);
//setThreshold(11, 255);
setOption("BlackBackground", true);
run("Convert to Mask");
run("Set Measurements...", "area mean standard min median limit display redirect=None decimal=2");
run("Analyze Particles...", "size=100-Infinity circularity=0.01-1.00 display exclude clear include summarize add");

The summary table gives me these numbers for the image attached
ARPE19IFNy-4dpi-Scene-01-B1-B01-Stitching-01-Image Export-01_c1.tif 265 1877700.887 7085.664 9.311 250.426 255
which would mean the average size of the plaques in this image is 7085 µm2?

Thank you for your help!

Marlies

ARPE19IFNy-4dpi-Scene-01-B1-B01-Stitching-01-Image Export-01_c1.tif (6.6 MB)

Hi Marlies,
Measuring the distance in the ZEN software will only work if the ZEN software has been calibrated. In your screen shot it shows in the status bar a scale of 1 px/px. It is a long time since I used the ZEN software, but this suggests to me that this is not properly calibrated - I would expect something like x px/um. It is also unusual that it would be exactly 1 pixel per um.
The best way of checking would be to take an image of an object with a known distance (e.g. a calibration slide) and then measure the length of the known object on the image (if you don’t have a calibration guide, but have a good idea how big the structures that you image are, you could use this as a rough guide).
Also, when you set the scale in ImageJ as you have done (run(“Set Scale…”, “distance=5000.4052 known=6455.941 unit=inch”)) , you actually say that 5000 pixels are equal to 6455 inches, which would mean that 1 pixel equals 6455/5000 inches, which clearly doesn’t make sense. It also means that your measurements are in inches as this is the unit you defined. Depending on your camera, etc I would expect that the calibration is somewhere around 1-2 um/pixel for a 5x objective.
There are various tutorials describing the spatial calibration process, e.g. Spatial Calibration - ImageJ.
Hope this helps,
Volko

2 Likes

Hi @Volko
The ZEN software gives these dimensions 1,29µm/pixel could thus be correct?

Using this command would work for all images (taken with same objective of course)?

run("Set Scale...", "distance=100 known=129 unit=µm global");

Entire macro

title = getTitle();
run("8-bit");
setAutoThreshold("Default dark no-reset");
//run("Threshold..."); 
setThreshold(11, 255);
//setThreshold(11, 255);
run("Convert to Mask");
run("Set Scale...", "distance=100 known=129 unit=µm global");
run("Set Measurements...", "area mean standard min median display redirect=None decimal=3");
run("Analyze Particles...", "size=150-Infinity circularity=0.01-1.00 display exclude clear include summarize add");

This sounds about right and is in the range of values I would expect for a 5x objective. Once you have applied this calibration to your images, you can use the line selection tool to measure a cell/structure of interest and see whether the length measurements in the ImageJ status bar look appropriate.
Good luck,
Volko

1 Like