How can I measure track speed with Fiji / TrackMate?

Good evening everyone,

I want to analyze the speed of Paramecium caudatum cells under various conditions. I use Fiji and the plugin TrackMate (more details: LoG detector, Simple LAP tracker). After having solved the usual problems, the program produces a lot of data after the Analysis step. One file contains among other data the “TRACK_MEAN_SPEED” which to me sounds like the speed of a given track. However, there is an inverse relation between the speed and the number of analyzed spots / track duration. So the longer the track duration, the lower the speed - which doesn’t make any sense. If you have a solution / advice, I would very much appreciate it. Thanks!

Hi @Lerchl

Indeed this correlation is unexpected. The TRACK_MEAN_SPEED is what you are looking for and returns what you expect.

A lead to pursue would be to check whether the tracking processed robustly and that the tracks are reliable and follow the real particles (no jumps to another particle, no breaks, no linking to spurious detection).

Could you post here a capture of the track overlay and visually inspect that the tracks are correct?

Hi @tinevez,
Thanks for the reply. Since I am not an expert using Fiji, I most likely did mistakes. Tracks Exp24.
I have recorded a total of 8398 frames (0.2 sec interval) which corresponds to 28 minutes. To save disk space I used a 3x3 bin reduction. The file size was 1.8 GB, the resolution 340 x 340 pixel. The resulting track overlay is shown above. I did not yet visually inspect that the tracks are correct.
May be I have to analyze first a small fraction of the frames to check for correctness… Any suggestion is more than welcome.

Ok given the density of tracks they are most likely wrong.

I suspect that most of the spots detected are spurious ones that are then built in tracks with random velocities. You must make sure that only cells are detected, not noise.

Can you check the tracking data?

Ok, the new analysis with some 500 frames looks much better. I still have to figure out the best parameters, but the data make much more sense now. The other problem I face now is how to load the data into Excel since there is a problem with the decimal point and the comma… But I’m sure to fix that problem, too. Thanks!

1 Like