Counting overlapping cells from different slices in confocal images

I have a stack of confocal images on a fly brain (1200x1200x720). The nucleus of each glia cell was marked with a cell type-specific gene tagged with GFP. I want to count how many glia cells is present in the fly brain.

To do this, I have used functions in ilastik to segment and identify cells in each layer. Next, I used the tracking function to merge the same cell spanning multiple layers. However, this function gave me a wrong estimate. I suspect it is designed to track cell/flies moving around with a continuous trajectory.

Any recommendations for merging cells spanning multiple layers in confocal images? Thanks!

ZZ

Don’t import the images as single slices.
Ilastik can work on volume data directly.

Turn the slices into a single volume and save as tiff or hdf5 and import the whole volume.
Edit:
This may be even easier if you use import single volume from sequence when adding the dataset.

1 Like

I second that. Do pixel classification on the 3D data set and then maybe Object Classification to clean it up.