Problem with neighbor counting when excluding borders


#1

I am trying to analyze epithelial cell morphology including count, area, shape and number of neighbors (image 1). When i input a section of the epithelial sheet into a cellprofiler, i currently seem to have two options either: 1 include images touching the border, which will provide me with data on these partial cells which are not fully in the picture and therefore have inaccurate measurements of shape, size and number of neighbors, or 2. i can discard images touching the border, which will provide me with the the number of cells that i want and an accurate count for shape and size, but will inaccurately count the number of neighbors for cells adjacent to the border cells, as these border cells essentially appear as One large cell (image 2)

Is there anyway that I can exclude border cells but still get accurate measurements for the number of neighbors statistic?

One idea that I’m working on, but am not sure how well it will work, involves overlapping the coordinates given in the output excel (defaultOUT.mat_id_objects) sheet from one run including the border objects with a run excluding the border objects, and then discarding any measurements given for coordinates not included in the run which discarded border objects. is this the only/best way to approach this problem? got any ideas?

Thanks a lot,
Nupur

Image 1- input image
Image 2- output image (mo_neighbors) excluding border cells.





#2

Hi Nupur,

I was just about to refer you to this page, but it seems that you already found it.
-Mark