Quantify htt aggregates


#1

Hello,

I would like some help modifying the pipeline I have created to quantify htt aggregates inside the Nucleus. I am interested in getting the number of those aggregates inside the nucleus, the area they occupy as well as the area of the Nucleus surrounding them. I have created a pipeline that recognize the secondary object (Nucleus) around the aggregates (Primary) but it is failing to correctly recognize the aggregates that are within the nucleus. We used to hand trace those images but that is a very time consuming and we want to switch to a more automated method. Any recommendations or help would be greatly appreciated

Here is my current pipeline
Htt aggregates pipeline.cpproj (415.0 KB)

Below is a sample picture I am working on

Thanks a lot in advance,


#2

Hello there,

You were close !
I changed a bit your pipeline, the main modification is:

  • In ColorToGray, instead of using “Combine” , use “Split”. It will give you a better separation between aggregates and the nuclei.
  • I’m glad you used the trick ImageMath to subtract the background from Inverted images, good job !
  • I’m also glad you used EnhanceOrSuppressFeature with Speckles.
  • You can identify aggregates and nuclei both as primary objects and then finally relate them. You can see here, when the aggregate find its correct “parent” it will happily change the color to match up with its dad/mom.

    Htt aggregates pipeline_modified.cpproj (645.2 KB)

Good luck.


#3

Thank you so much for your great help. I just deleted one of the filters, used the Automatic threshold strategy, and the shape method to distinguish clumped and it worked beautifully !!! I appreciate the time you put in this ! Thanks a lot for your quick responses and the great help again !


#4

I am still trying to optimize my pipeline for the aggregates and I am having some difficulties improving the detection accuracy. my current pipeline is missing some clear aggregates sometimes and it is selecting a lot of them in one nucleus other times when it is clear that there is only one aggregates. Any ideas on how to improve my pipeline would be greatly appreciated.

here is my current pipeline aggregates (63x).cpproj (420.5 KB)


#5

This is a picture I am using


#6

Hi,

If you have over-segmentation problem, you can try to hit “No” at :
“Automatically calculate size of smoothing filter for declumping”
“Automatically calculate minimum allowed distance between local maxima”

Then increase the values in :

  • “Size of smoothing filter”
  • “Suppress local maxima that are closer than this minimum allowed distance”

The values to input there should somewhat reflects the usual size of one (01) aggregate.

Hope that helps.