Histology (H&E) muscle fibers quantitative analysis

Hi everybody,

After some weeks playing with CellProfiler in order to find a pipeline adapted to analyze my histology samples, I hit a wall.

I am conducting a research project to understand skeletal muscle growth regulation in fish. For this project, I have a large quantity of histology slides to analyze (Hematoxylin and Eosin stain). I would like to record the number of muscle fibers and record their area/volume/diameter for each pictures in order to see if treated fish have developed an hyperplasia (increase in the number of muscle fibers) or/and an hypertrophy (increase in size) of the muscle fibers.

To do that, I used and mixed different pipelines and comments that I found in the forum and try to improve it for my pictures, however, it is not a success yet.

My objectifs:
1- Removing the cells that are not full around the edge
2- Improve the detection of the cells and fill the inside of cells to retrieve the true volume of each fiber
3- Classify the cells and determine how many cells per range of sizes (i.e # of small, # of medium and # of large fibers)

I linked to this post some pictures and my current pipeline.

Thank you in advance for any help and feedback



Pipeline_Histology_MuscleF.cpproj (392 KB)



Hi,

Please try the attached pipeline. Histology analysis is always tricky and not very robust, but this is hopefully an improvement.

In particular:

  • UnmixColors: If you have color histology images, you should at least try this module
  • Morph, Smooth, EnhanceEdges: These preprocessing steps are all trying to enhance the borders between the large regions. Feel free to add/remove or changes scales as you see fit
  • ImageMath > Invert: Simply used because some downstream steps need to have bright foreground, which may be the borders, or the red regions, depending.
  • IdentifyPrimaryObjects: It uses a strict cutoff of object diameters which may need to be adjusted. It uses “Shape” as a declumping method, which assumes concave objects and the fiber clumps mostly are. I have manually adjusted the declumping parameters and you will likely have to adjust them further. The Fill Holes option can also have an effect here.

Hope this helps!
David
DL_H_and_E.cppipe (9.58 KB)

Hi all,

Sorry for the late response. Thank you David for your help! Your pipeline seems much more logic than mine.
I will play with it and add the measurement tools on the pipeline this week.
I hope it will work well for all the pictures.

I will keep you updated :wink:

Thank you again

Kind regards,

Adrien Marc

CellP1.zip (1.0 MB)
CellP2.zip (492.2 KB)
Hello;
I work on muscles cut transeversally.
I made identical cuts on different blades, after I made the staining of each blade by a specific dye.
In the first coloration we see all the cells in red, from this coloration I want to calculate the total number, the average diameter, the perimeter, the Feret diameter, the area occupied by the cells, the percentage of fibers that have A small, intermediate and large size.
2- the second coloring allows us to see three types of fibers: fibers with a dark blue color, light blue and white.
I want to compute from these images the total number, the surface, the perimeter, the average diameter of Feret of each type.
3- the third color reveals two types of fibers: black and white.
I want to calculate from these images the total number, the diameter, the perimeter, the average Feret diameter of the black fibers.
The white fibers I will calculate them from the 2 other images.
Can you send me pipelines that process these three types images independently or at the same time.
thank you in advance.