Quantifying apoptotic cells

mpar

Good morning,

I am trying to quantify cells in mouse hippocampus and am struggling to segment the images. Specifically, I need to count apoptotic cells in the CA3 region. I have Nissl and H&E stains at my disposal. Here is an example of the images I'm working with. I used Paint to indicate the region I would like to measure. Is it possible to count the total number of cells in this area and then add a second object type for the cells of interest? I added a few annotations to give an idea of which ones I'm talking about.

As you can see on the Image Strip, each of these slide images has three sections on it, and I would like to perform the same count on each.  

I have used ImagePro in the past but it has been a long time.

Any ideas? Unfortunately, I don't have any other stains to work with yet. I know this would be much easier with a TUNEL stain, for example. I can also upload a crop of the original .tif image if that would be beneficial. 

Thanks so much!

YuriG

Hi mpar,

This type of measurements is related to parent/child category. There are several ways to do it in Image-Pro, I will explain one of them.
First of all you should find a way to segment CA3 areas and apoptotic cells. You can use different Smart segmentation recipes to segment these 2 types (tip: use larger Size parameter for CA3 areas and smaller for cells).
If you manage to segment these 2 types of objects, then use the following steps to get cells count per area:
1. Load segmentation recipe for CA3 areas and click Count. (you should end up with 3 objects)
2. Open Measurements Options dialog and select "Preserve Existing Data" and "Use as Outlines" options.
3. Load recipe for segmenting apoptotic cells and click Count. It will Count cells only inside CA3 areas.
4. Add measurement "Object: Parent" (in the Types dialog), if you don't have it yet.

The Measurements data table will show the list of all cells where Parent field points to one of CA3 cells. Use "Group by Parent" menu item (in the Grouping button dropdown of the Measurements table) to group all cells by CA3 parents, the Totals field can be set to Count to report the number of cells per parent.

Yuri 

mpar

Thank you so much, Yuri!

mpar

YuriG said:

Hi mpar,

This type of measurements is related to parent/child category. There are several ways to do it in Image-Pro, I will explain one of them.
First of all you should find a way to segment CA3 areas and apoptotic cells. You can use different Smart segmentation recipes to segment these 2 types (tip: use larger Size parameter for CA3 areas and smaller for cells).
If you manage to segment these 2 types of objects, then use the following steps to get cells count per area:
1. Load segmentation recipe for CA3 areas and click Count. (you should end up with 3 objects)
2. Open Measurements Options dialog and select "Preserve Existing Data" and "Use as Outlines" options.
3. Load recipe for segmenting apoptotic cells and click Count. It will Count cells only inside CA3 areas.
4. Add measurement "Object: Parent" (in the Types dialog), if you don't have it yet.

The Measurements data table will show the list of all cells where Parent field points to one of CA3 cells. Use "Group by Parent" menu item (in the Grouping button dropdown of the Measurements table) to group all cells by CA3 parents, the Totals field can be set to Count to report the number of cells per parent.

Yuri 

Good afternoon, Yuri. I have managed to segment the CA3 region (using the Magic Wand) and extract it as an ROI, but I am having trouble with the Smart segmentation. Specifically, I am not sure how to use the Size parameter, which seems like it would make this easier than trying to use the Magic Wand. 

What am I missing? Thanks for your help!

YuriG

Hi mpar,

I explained a general workflow of getting parent/child data. It's assumed that you can segment parent and child objects. If segmentation is the problem, it must be addressed first. Can you send a test image and the annotated image which shows what is a cell, so I can check if the segmentation is possible?

Yuri

mpar

Hi Yuri,

Yes, it is the segmentation that is giving me trouble. I think part of the issue I'm running into is that this stain is not ideal for what we are trying to measure. I'm working on a more appropriate stain for this which I think will make things much easier.

Here is a sample image with yellow annotation markers over the sort of cells we are trying to count.

mpar

Here is the same image without the annotations. I appreciate your help very much!

YuriG

Hi mpar,

It's not an easy sample. According your annotation the cells are bright objects. So you can use some filters to enhance the objects before segmentation. I used Band Pass filter with the parameters as on the screenshot and then executed Manual Threshold segmentation with auto-split, the option file SegmBandPassThreshold.iqo is attached. The result looks like this:

You can achieve similar results on original image using Smart segmentation with custom Band-Pass channel (attached SegmSmart.iqo):

I don't know if that is what you want. You can tweak the Filter Ranges to filter out unwanted objects.

Regards,

Yuri

mpar

Yuri,

This is so incredibly helpful and so appreciated. Thank you!

Is there a beginner-level resource available to explain the principles behind each of the filter types and how to best use them? I am a bit out of my depth here and would love to understand this properly. 

Thanks again.

YuriG

mpar,

You can find many useful tutorials and webinars in our media library : https://www.mediacy.com/imagepro/learn/media

Yuri