Splitting cells

[Adam]

Hello,

I have set splitThreshold = 0, but it still combines too many adjacent cells int a single cell. Splitting manually works fine. Is there another parameter I can adjust to simulate what I'm doing when I press the "split" button?

Thank you,

Adam

[Seán Murray]

Hi Adam,

first, have you split1=1 (for independent mode and the first frame of timelapses, otherwise the splitThreshold parameter has no effect)?

the program uses edge and detection algorithms after thresholding (and I think watershedding). http://microbetracker.org/help/helpValleyDetection.htm

Depending on the parameters for these algorithms, cells will be separated or not. Furthermore cells which are separated using these algorithms are not marked with ancestor/daughter information (because they could be side by side). Then, after the alignment step, if split1=1 and/or splitThreshold <1, the program will try to split cells.

So one approach is to choose the parameters such that side by side cells are identified and leave any remaining end-to-end cells for the splitting algorithm.

One annoying thing is that the program does not show you the cell outlines after this edge detection but before the time-intensive alignment step. The segmentation button just shows the mask and not the resultant outlines so one has to finish a complete analysis to see the effect of changing the edge detection parameters. And its more difficult to identify at which step problems are occurring.

Unfortunately, it seems Oufti (Microbetracker2) will not implement this. I might do it myself at some point. 

Sean

[Adam]

Hi Sean,

first, have you split1=1 (for independent mode and the first frame of timelapses, otherwise the splitThreshold parameter has no effect)?

Thanks for the help. I did not notice that split1 = 0 by default and that this overrides the splitThreshold value. Perhaps this should be made a little more clear in the parameter description. It is working much closer to what I expect now.

I have two remaining issues. First, is 'areaMin' just a suggestion? I have set it to 60 but consistently get cell outlines that are much smaller than this (~40). Second, I get a number of "ghost cells" that do not seem to be associated with anything in the phase image. I have attached an annotated screen shot.

If you want, I can repost as a new topic.

Best,
Adam

[Seán Murray]

Hi Adam,

best make a new topic as the answer might be useful for others.

Then I'll post the explanation and fix.

Sean