For an image analysis project I am working on together with a biologist, I am using the Haralicks feature “Cluster Prominence” to quantify the distribution of signal (see also here: First Order Statistical Analysis). This works really nicely, and with this metric I am able to distinguish well between cells in which the fluorescently labeled protein of interested is distributed evenly, and cells in which it tends to form punctae/spots/aggregates/dots.
What I struggle with is an EASY explanation of this metric, and whats going on under the hood when this is calculated., so that it can intuitively be understood.
The standard in the lab or maybe the field for this type of quantification has previously been “count the number of dots”, which is certainly easier to understand, but way less precise and suffers from bias when defining whats a “dot” and what isn’t.
Any further thoughts or ideas are much appreciated!