Skip to content, Skip to search

Changes

TrackMate Algorithms

3 bytes removed, 10:35, 16 January 2015
m
LAP trackers
The first one is simply a simplified version of the second: it has less settings and only deal with particle that do not divide nor merge, and ignores any feature penalty (see below).
All the linking costs for these two trackers are based on the particle-to-particle square distance. If this tracker had to be summarized in one sentence, it would be the following: <b>The SImple Simple LAP tracker and the LAP tracker are well suited for particle undergoing Brownian motion.</b> Of course, they will be fine for a non-Brownian motion as long as the particles are not too dense.
Particle-linking happens in two step: track segments creation from frame-to-frame particle linking, then track segments linking to achieve gap closing. The mathematical formulation used for both steps is linear assignment problem (LAP): a cost matrix is assembled contained all possible assignment costs. Actual assignments are retrieved by solving this matrix for minimal total cost. We describe first how cost matrices are arranged, then how individual costs are calculated.
# In <b>u-track</b>, merging and splitting of tracks are used to describe two particles that temporally overlap spatially. These events' costs are weighted by the two particle intensities to properly catch the apparent increase in intensity due to the overlap. In <b>TrackMate</b>, we use splitting events to describe cell divisions, as we developed it initially to deal with <i>C.elegans</i> lineages. However is seems than Jaqaman and colleagues used it the same way to investigate CD36 dissociation and re-association.
# In <b>TrackMate</b>, distance and time cutoffs are specified manually by the user. In <b>u-track</b> they are derived for each particle automatically, providing self adaptation.
 
 
 
=== Linear motion tracker. ===
Emailconfirmed, incoming, administrator, uploaders
1,396
edits