Quantification of Mitochondria without MitoTracker Staining

Hi All,

I have TEM images of mitochondria, and I am trying to quantify them based off several parameters like number, size, and circularity using ImageJ. I have been reading about the macro developed by Dagda (https://imagejdocu.tudor.lu/plugin/morphology/mitochondrial_morphology_macro_plug-in/start), and it seems promising. However, we have not used the MitoTracker stain for our images so the macro does not work properly. I don’t have much experience with using ImageJ, so would anyone be able to push me in the right direction? Thank you!


So… the first thing you need to do is segment your mitochondria - then you can get all the measures you need. Would you be able to share an original image file with us by uploading one here directly on the forum or via a link to a file sharing site (e.g. Dropbox)? Once you segment - you can calculate size/shape parameters of your objects using tools such as: Measure via Set Measurements or Shape Filter.

Here are some helpful links to get you started on ImageJ and then segmentation:

1 Like

Hi, thank you for your response! Attached is an original file image. Is the purpose of the segmentation to identify what is a mitochondria and what is not? How exactly does it work? sample mito.tif (11.4 MB)

In your case - yes. Segmentation is the process of partitioning an image into multiple pieces i.e. object versus background or mitochondria versus everything-else…

Go through the links I provided you above - especially the Segmentation workshop linked above. TWS (also linked above) will be the most applicable tool for you to use given your dataset. The workshop goes over a quick how-to in using that plugin…

Great, thank you! I forgot that Fiji is slightly different from ImageJ in terms of plugins. So I’m playing around with the segmentation tool right now. Thanks for the help!

1 Like

So I’ve gotten the segmentation to identify most of the mitochondria, but it’s also picking up background. Is there a way to make the two more distinct? Additionally, is there a method to be able to count the mitochondria after segmenting them? Thank you!


You just have to iteratively train the classifier… adding more labels (ROIs) to train on which pixels should be considered mitochondria and which shouldn’t. So in that case - add more labels to the not-mitochondria class.

And for segmenting them there after… you can generate results - either labels or probability maps - and then use built-in segmentation tools within ImageJ. For that - go through the links I gave you above. Like I said before… especially the segmentation workshop. All the information you need can be found in those links.