Counting DAPI Nuclei with Chromatin Speckles

We are attempting to run cell counts on these DAPI-stained nuclei. However, all traditional threshold/masking techniques I’ve tried in FIJI haven’t worked, as FIJI is confused by the strong chromatin speckles in each cell. Does anyone have any idea for how to go about counting cells like these?

As an added bit of complexity, we’re only interested in the cells between the orange lines, but I’m less concerned about solving that issue now.

region.tiff (14.0 MB)


Ouch, that combines two of the bigger IF problems I have run into :slight_smile: Bright spots and elongated overlapping nuclei. I haven’t really taken the time to try it, but I have wondered if truncating the brightest spots to a set value might help reduce their disruption (without the blur from a median filter). Best of luck though :slight_smile:

Apologies if you already have tried this. But I have a lot of success counting densely packed nuclei using the ‘Trainable Weka segmentation’ in FIJI under plugins->segmentation. If your original data is in a z-stack, I found the 3D segmentation also located there to be even better.

With your orange lines, the simplest way is to draw those lines as selections in imageJ, adding them to a ROI manager, then Edit->Clear outside (for the outer ring) and Edit->Clear (for the inner ring)

Hope this is of some help and good luck!