Hi, I have not used CellProfiler much before. However I have managed to find the objects I want in my image (Cells and nuclei). Now I want to train the classifier in CellProfilerAnalyst to count the cells which have more than three nuceli.
I have struggled with getting CPA to work as I would like. In CP I have done a lot of measurements of my objects. I also here make the properties file for CPA and the database file. When I open CPA I chose the properties file that I have just created. When I open the classifier tool and select Fetch 20 random objects, it does fetch 20 objects. However, all of the images of the individual fetched objects does not appear in the unclassified area. They only appear as black squares. When I try to open the whole image for that object I get an error message.
When I check the table they ask for, and look for coordinated to the Cell locations, I see that some of the does not have coordinates:
I don’t understand why CP can’t find a location for some of the cell objects?
I also see that these cells (without location) has no children (Nuclei), described as None. However another cell with location, has 0 children :
When I check the PerObject table I see that I have many cell objects without location. In total I have 2157 cell objects, but only 1507 of them have a location.
Is there a way to remove these cell objects without location? Or is there a way to find their location? When I get the coordinates for the cell objects into CPA, I can distinguish the objects myself with the classifier tool, and put the nuclei-less objects in the negative area Might it be that I need to improve IdentifyPrimaryObjects of the Cells in CP?
Thank you for advise! Let me know if you want me to send some files or images