TrackObject problem, same label assigned to multiple objects

Dear,

I’m trying to track cells using TrackObject module. I have questions about overlap and LAP method. I’m using r11710 version with Mac OS X 10.7.

I use Overlap method first. I occasionally see more than two objects has identical label number. This happens when those objects splits from a single parent. This issue had already been posted by tischer on Nov 19, 2008.

My first question is whether there is any solution found for the issue since then?

I also tried LAP method. I had another problem. It seems TrackObjects_Label in output .csv file doesn’t match with object numbers in images saved with SaveImages. Also TrackObjects_Label in .csv assigns identical label number to different objects in same image. I wonder this is a known bug or just my erroneous pipeline? If you know any solution for this, please teach me.

I attaches my pipeline and output .csv files. I wanted to upload my images generated by SavedImages, but .mov or .avi file are not allowed. If you need them as well, please tell me how to do it.
DefaultOUT_RedCells.csv (3.12 MB)
DefaultOUT.mat (3.35 MB)

Hi,

[quote=“hshigeno”]I use Overlap method first. I occasionally see more than two objects has identical label number. This happens when those objects splits from a single parent. This issue had already been posted by tischer on Nov 19, 2008.

My first question is whether there is any solution found for the issue since then?[/quote]

The general issue of how to best handle progeny during tracking (e.g., do they receive altogether new labels from the parent, or a continuation of the parents label?) is still an open question for all of the tracking methods. An more in-depth explnation of this issue is given here by our software engineer who coded the module.

If you are using the 2nd phase of LAP, then this is to be expected. The images saved by SaveImages reflect that of the 1st phase only, since the 2nd phase depends upon initial tracks created by the 1st phase the and runs only at the end of pipeline. We unfortunately do not currently have a workaround that allows for visualization of the labeled objects post-run, other than by using scripts in 3rd-party software (e.g., MATLAB) that read the output .mat file.

The above link also addresses this question.

Regards,
-Mark

I read the above posts, that partly addresses my question, but does not answer it.

I also read in the manual (cellprofiler.org/CPmanual/TrackObjects.html):
Label: Each tracked object is assigned a unique identifier (label).** Results of splits** or merges are seen as new objects and assigned a new label.
ParentImageNumber, ParentObjectNumber: The ImageNumber and ObjectNumber of the parent object in the prior frame. For a split, each child object will have the label of the object it split from.

Isn’t that contradictory, or am I just confused? :blush:

In my case, the tracked objects (same pipeline as the demo one for tracking), when they split, keep the parent number, which is fine with me. However, I would like to individually identify each object for later processing. I haven’t seen a way/parameter that would allows that. Did I miss something here?

Hi there,
I posted the above question in May 2014, but haven’t seen an answer since. Could someone try to explain to me what “Results of splits or merges are seen as new objects and assigned a new label” really means, since this new label appears to be the parent’s label, which is not a “new” label.
Is there at present time a parameter that would allow to individually identify each object following cell division? If yes, which one is it?
Thanks in advance for your input.
Philippe

Hi Philippe,

Sorry to say that the quoted statement is simply wrong, a mistake on our part. We have changed the text to read “Child objects resulting from a split or merge are assigned the label of the ancestor,” which will be documented in the next release. We have it on our to-do list to provide the user with the option of children of a split/merge inheriting the ancestor label or getting a new label (github.com/CellProfiler/CellProfiler/issues/642) but we don’t have a timeline when then this will be done.

Regards,
-Mark