Kaitlyn’s notebook: Quantifying geoduck histology

These are possible measurements that can be taken on the geoduck histology slides based on tools I’ve found:


  • pixel classification: acini vs connective tissue
  • average acinus size (measure widest portion of 25 acini/individual)


  • oocyte length
    • measure only round eggs with visible nucleus
  • average follicle size (measure widest potion of 25 follicles/individual)

Unfortunately I can’t use a pixel classifier for the females because the oocytes and connective tissue stain is very similar in color and there are significant white space between the tissue and in the nucleus. 

I have been using the Padilla-Gamiño micrscope (Nikon Eclipse Ni-U with the Nikon Elements BR [basic research] program). I have been exploring the program to find features useful to our histology slides. I can get the % area using the pixel classifier.



Acini are deeply stained. Using the pixel classifier I can make these pixels red. The percent area for each phase (pixel stain) is shown to the right.


Connective tissue is shown with the green pixels.


A possible third phase in blue…

Alternatively, I can do just two phases-


this is using the manual method (the bayes method has you pick sample pixels for each phase rather than showing a graph for the wavelengths).

However, the inability to remove the phase overlapping on the image makes it very difficult to find another spot to measure on the slide, and I surmise that at least 3-4 measurements throughout the slide should be taken to accurately determine the % area of the acini.

Other possible option are thresholding in Fiji (Fiji is just ImageJ [but updated]), color deconvolution in ImageJ (but I have not been able to find a way for the program to give me % area), or pixel classifier qith Qupath (a new addition to the program that still has soem kinks and I haven’t been able to use yet).

It is easy to take measurments and images with a scale burned onto the image so measuring oocytes, collicles and acini using either this program or ImageJ is easy! I am working on females first, and during any down time, i.e., someone else needs to use the scope, I am trying to learn how to sue Fiji, Qupath, or the Element BR program.