Segmentation of foci#

In this second part of our workflow, we want to detect foci within the nuclei as individual objects. Foci are much smaller than nuclei so their brightness is more heterogeneous in nature. Additionally, they belong to nuclei that can be brighter or darker, so this segmentation proves even more challenging than the segmentation of the nuclei themselves as they all share a relatively dark (i.e. low signal) background.

Analysis pipeline steps#

Pre-processing steps#

Using the same data we did in the nuclei steps, load 2D test data into napari:

  • Load the plugin Napari-segment-blobs-and-things-with-membranes and choose the Gaussian Laplace filter.

  • Choose the image layer

  • Click run. Note that the image is inverted, i.e. everyting that was dark is now light and everything that was light is now dark. We will use this information to create our analysis pipeline.

Gaussian LaPlace filtered image

Segmentation steps#

Load the PartSeg plugin and choose the ROI analysis extraction option. Set the parameters as follows:

  • Target Layer name: foci or a name of your choosing

  • Choose Upper threshold from the dropdown below the run button

  • Mask: for this example choose the previously created nuclei mask

  • Channel: gaussian_laplace result

  • Filter: none (this image is already filtered)

  • Threshold: Manual

  • Threshold: -2

  • Minimum size (px): 3

  • Connect only sides: should be checked

ROI analysis extration option