Manual post-processing of identified objects

Dear CellProfiler users,

my name is Marco and I’m a newbie cellprofiler user, I’ve found problem in automatic identification of secondary objects. I’ve got images of confluent elongated adherent cells, the provided tool is quite precise in identification but sometime, due the particular cell morphology, it’s not able to correctly identify some cells. I’ve tried different threshold and post processing but things got just worse. Could you help me? Could be performed any manual modification of identified objects?

my pipeline (basic)

LoadImages
ColorToGray
…different modules as enhanceEdges or Feature ect
PrimaryObjectId (for nuclei it’s always precise for correct tyoical diameter
SecondaryObjectId (quite precise with watershed-gradient, but I’d like to modify not correctly identified entities)
MeasureObjectSize
Save

Images are attached

BestRegards




Hi Marco,

I have to say, the cell identification doesn’t look too bad! :smiley: But if you’re really determined to correct the remaining errors, CellProfiler 2.0 doesn’t have a method to do so. However, CellProfiler 2.1 (to be released soon) does have manual object editing capability; you’ll be able to select objects points along the boundary and modify them as needed. So stay tuned!
-Mark

And just one other comment after looking closely at your images – it is the Identify Primary Objects step that is failing (though, it is mostly good, as Mark says). In the upper error, one IDPrimary nucleus is not missing, and in the lower error, the two nuclei are erroneously merged. I would focus your efforts on nucleus ID and that is usually best done in the staining/wet protocol since DAPI or Hoechst are usually very distinct.

Cheers,
David