How to score image phenotypes?

Hi everyone,

I would like to score tissue phenotypes in an image in a semi-automatic format. I tried using morphologic analysis to detect different shapes, but it hasn’t been really successful, so I think I’m just going to do it manually.

I have modified a macro on here to split my tif into 50 smaller rectangles (e.g. image1-tile001, image4-tile048, etc.). Now I have a folder containing the files I need. I would like to be able to:

  1. select a random image without repeat
  2. display that image on screen
  3. press a key to assign phenotype (e.g. 1=thick, 2=mixed, 3=sparse, 4=nucleus, 5=n/a)
  4. record the score with the image label
  5. move to next image
  6. at end of folder, show report with each image and percent of tiles for each image (e.g. “image 1: 20% class 1, 20% class 2, 10% class 3, 50% class 4, 3 tiles not scored”)
  7. profit?

I’m thinking the best way to do this is probably with python or some macro within fiji. It makes me think of Anki flashcards I used back in med school. Am I asking in the right place?

Thanks in advance for your help!

Hi there,
I recently proposed a set of fiji plugins for this kind of phenotypic annotations.
See Fiji plugins for qualitative image annotations:… | F1000Research and the youtube tutorials.

If you are assigning a single category per image, then you can use the F keys (instead of the buttons) to assign the categories with the “Single category (button)” plugin.

It can open images in a directory one after the other in “directory” browsing mode.
It does not do the random image opening, but if you find a way to do it in Fiji, then you can just untick the “Auto next” option in the plugin and use the plugins for annotations only.

Not sure what you mean by score in your step 4 though. My plugins allow the assignment of one or multiple categories for each image but no probability/score is associated.


Wow, thanks for sharing that! I’ll check it out. it definitely sounds the like the right function. As for randomization, I can just randomly sort the files in the directory first.

By 4, I just meant the assigned phenotype. (Scored as class a, b, c, etc.).

This is exactly the thing I was looking for. Thank you so much! I hope my PI likes it :grimacing: