Roi tracking in TrackMate (not PointRoi)

Hello everybody,

I’m using trackmate to track objects from a binary timelapse mask.

To detect and track, I’m using trackmate and the BinaryDetector: a trackmate extension which uses IJ’s ParticleAnalyzer. This part is working fine.

Here’s the problem : I want to do some analysis on the original ROIs detected by the Particle Analyzer, but they’ve been lost into TrackMate, which handles each spot as a blob with double properties, if I get it right.

I’ve tried to retrieve these ROIs by using the wand selection in the original mask image, taking the coordinates of the spots as the wand location but… I have elongated non convex objects, and thus I miss objects in many timepoints:


Any idea how to solve this ? I’m currently trying to find a way to create a linked map between spots and objects, directly modifying the BinaryDetector code. It’s tedious and not obvious. Maybe there’s a better option ? An already existing option ? A clever trick ?

Thanks for your help and suggestions!

1 Like

Traditionally, the columns XStart and YStart are the ones to be used to reconstruct binary particles from a single coordinate. These are recorded when the Record starts options is checked in the particle analyzer. See also the documentation:

Record Starts This option allows plugins and macros to recreate particle outlines using the doWand(x,y) macro function.

I guess you should be able to record these by handing your own options to the constructor of BinaryDetector

I didn’t use BinaryDetector myself yet, but from glancing quickly at the source code, I can see a potential problem: it seems to rely on the presence of the columns Feret, MinFeret, XM, YM and Max, but doesn’t make sure the measurements given to the constructor fulfill those requirements:

From my understanding, you should be able to keep using the centers of mass for tracking, but adding XStart and YStart as additional spot features.

1 Like

I wasn t aware of XStart and YStart.

Sounds like the way to go. Thanks!