Identifying vasculature (DAB)

Hi there,

I am interested in determining the number and area of vasculature on CD31+ stained tissue (DAB) (Image 1). Due to the fact that the vasculature is elliptical and ring-shaped, I am having trouble accurately identifying them. As shown by the OverlayOutlines image output, it seems as though a greater area of the vasculature is being captured (Image 2). I have tried the EnhanceOrSuppressFeatures module in order to enhance speckles, but I found that the output was non-specific. Furthermore, I have tried modifying the thresholding methods but no luck. With variable intensities of the DAB stain, I used FilterObjects to only keep cells that reach a minimum intensity. I tried using IdentifySecondaryObjects in combination with the IdentifyPrimaryObjects module but it wasn’t able to recognize many of the smaller or elliptical vasculature.

I’m wondering if anyone has any ideas about how to proceed in accurately identifying the vasculature? My pipeline is also attached in the Dropbox folder.

Also, as the starting images, I deconvolute an image containing DAB and hematoxylin (image 3) so that the DAB becomes separated from the hematoxylin (Image 1). I’m wondering if there is a module on CellProfiler that can separate an image into its stains or would that be the UnmixColours module?

Any help would be much appreciated.


PS: To the CellProfiler team - the new forum looks great!