I’m currently struggling to achieve a good segmentation of focal structures in 3D.
This folder contains an image stack (the “raw” file) and the same file from which the background has been subtracted by applying the Differences of Gaussian filter. This first step gave good results in terms of removing the background while retaining also signals with lower intensities which is important for my subsequent analysis.
As you can see, the signals present in this stack are quite different from each other, i.e. some are clearly individual focal signals, but most of them are somewhat connected by either lower intensity voxels that indicate that such objects actually consist of multiple focal structures and yet others are more closely connected to each other. The latter ones we cannot resolve with the resolution of the system used to acquire this data.
What I would like to achieve is a segmentation of all signals (which is relatively easy with the background subtracted file with background values set to “0”) with an additional separation of those that are only connected by small “intensity bridges”. I tried many different possibilities, including either initial histogram normalization or even equalization to bring up low-intensity signals followed by detection of maxima and segmentation of the underlying structures with e.g. the MorpholibJ or the Find Foci plugins but without success so far. I really have difficulties in putting together a workflow that - in combination - gives good results…
Can someone with more experience maybe help me out here? I appreciate any help!
Best wishes and thanks,