Tracking label



Hi all,

are you sure the labeling is unique, as stated here:
“Each tracked object is assigned a unique identifier (label). Results of splits or merges are seen as new objects and assigned a new label.”

it doesn’t look so, in the attached picture…

thanks for helping,


Hi JR,

Yes, you’re absolutely right, the documentation is blatantly wrong :smiley:

We’ve updated the docs in our source code, which has not been officially released as yet. There are pros and cons to creating a new label for each daughter objects, but it’s still an open question.



Hi Mark,

I am new in biology and I recently started to deal with segmentation and tracking…
I think your output is better than few others, I’ve seen so far, because you still have the column :
that gives you information about his history, so, in principle one could reconstruct the “phylogeny”…
( I guess, with something like, than you might want to implement already in your code and print the phylogenetic tree…that would be cool)

…given that one has different label for the dauthers…so, that’s is important…

so, at the current status, is it possible to obtain a time-series dataset, form your software?
(a time-series, for some channel - signal, size, etc…- for each object, defined as one of the 2 blobs appearing between two cell division)



Hi JR,

At this point, you would have to parse out the spreadsheet to obtain the track for a given cell/object. So for example, you’d have to start at the beginning of the movie, find the object label and then follow the parent column to “connect the dots”. CellProfiler doesn’t do this natively, so you’d have to use third-part software.

Another feature that might be useful for this purpose is the so-called “relationship table”; you can find this in ExportToSpreadsheet as one of the default spreadsheets created, or by selecting “No” for “Export all measurement types?” and then select “Object relationships”. Or in ExportToDatabase, selecting “yes” to “Export object relationships”. In either case, the table created contains the parent/child relationships for all objects, including tracked objects, one object per row. This is a more robust way to keep track of them, because in the per-object table, events such as merges can’t be recorded (since an object would then have two parents, but space for only one of them in the ParentImageNumber/ParentObjectNumber column), whereas the relationship table can do so.



This is one of our main issues with cellprofiler… Do you know which third-party software can do this?


For visualization, you could try . But for quantitative analysis, I don’t know offhand if any tools exist for CellProfiler tracking data.