Ilastik tracking file column definitions

Hi, I was thinking if there is description available for the act of the column headings in the csv files that comes out of the plugin in tracking workflow. I am most interested in what Average_branch_length and Number_of_branches mean. Any help pointing in this direction is greatly appreciated!

Hi @glenzos,

welcome to the Forum! ilastik uses the vigra library to compute object features. Relevant parts of the vigra documentation in the function extractSkeletonFeatures; specifically:

  • ‘BranchCount’ (= Number_of_branches): the number of skeleton branches (i.e. end points after pruning)
  • ‘AverageLength’ (= Average_branch_length): the average length of the skeleton’s branches after pruning

Hi @k-dominik! Super thanks! ilastik already had made life easier and this makes it much better!

1 Like

Another related question to the csv file output. Is there a possibility to derive the time/number of frames required for one object to split into two from the tracking output file? or based on their lineage/parent ids?

the trackIds are unique per object between events (appearance, disappearance, division) and carry through mergers. What it , unfortunately, doesn’t tell you, is how long the actual division takes. For ilastik a division event is characterized by two subsequent time-frames where in the first time frame you have a single object, and in the second, there are two.

So, a way to maybe indirectly measure it would be to train an additional object classification project. If the division is visually perceptible, you can train a classifier that classifiers objects into dividing and not dividing in the “real” sense (not the “artificial” definition used in tracking). If you use the same binary mask, or the same probability map with the same thresholding settings, you can afterwards match objects based on their IDs. If you match them with the tracking you can only look at those that are actually dividing (maybe to filter out false positives)…