Using TrackMate on RFP and GFP expressing cells

Hi all,

I’ll preface my post by saying I am a FIJI novice–which is why I’m reaching out. I’m interested in tracking cells for a few .avi files I have. However, I have been unable to make the trackmate plugin properly identify my cells. I’ve tried LoG/DoG and many threshold settings, but trackmate consistently misidentifies cells. I’m starting to believe its an importation/file type error or perhaps its something simple that I’m overlooking. If anyone has experience tracking colored cells I’d really appreciate your advice. I realize its likely I haven’t provided enough information in my post to diagnose a problem, so please feel free to ask me for more information as well. Thanks!

Could you give us a sample image? Could you try if you can identify them by process -> find maxima?
If that is the case, then you may try to use the find maxima detector for trackmate:

Hi twagner,

Thanks for the response! I’ve attached a sample image. A few more things I’ve also noticed:

-Trackmate will only function if I change the image type to an 8-bit color image.

-When initially clicking on the trackmate plugin once I’ve imported my avi file and changed the image type to 8-bit color, the trackmate GUI says “Target: RFP.” Do we have any idea why trackmate defaults to only RFP? I’m interested in tracking both GFP and RFP cells.

Thanks again for all of your help!


I’ve tried the find maxima and messed with the noise tolerance for a while and was able to come up with this–is there any way to get the program to be more specific? I.e. not have multiple “maxima” in one cell? Thanks!

Hi @bgolson09 ,

I would try to analyse the green and the red cells separately. The following procedure is for the green cells.

  1. Split the channels Image -> Color -> Split channels
  2. Subtract the red channel from the green ( Image -> Process -> Image calculator). Let the result be the image A
  3. Now subtract from A the blue channel. Let this result be the image B. B only contains the green cells.
  4. Apply a mean filter on B with size 4
  5. Now apply the find maxima with tolerance 50
    This is the result:

With a tolerance of 30 the results seems to be even better:


I’ve changed the procedure (basicly I do not subtract the blue channel) so that you also get the low contrast cells:

  1. Split the channels Image -> Color -> Split channels
  2. Subtract the red channel from the green ( Image -> Process -> Image calculator). Let the result be the image A
  3. Apply a mean filter on A with size 4
  4. Now apply the find maxima with tolerance 30
    This is the result:

Hi @bgolson09

I would second @twagner on separating the channel but I would take it further.

Do NOT try to work on the AVI file. Go back to the RAW files that were generated by the microscope, and analyze each channel separately in its native format. Here you just add major complexity and problems using an AVI which compressed, merged and do unspeakable things to your scientific data. The AVI is used for communication normally. TrackMate works best with raw data.


@Twanger thank you so much for your help! I’m really making progress–the find maxima detector for trackmate worked fantastic. I’ll post an update later as I continue to work on the data!

Hi @tinevez,

Yes–absolutely. I just spoke with someone at my institution and they said the same thing. Thanks for your help!

@twanger Just wanted to say thank you again for all of your help. The end-product of my tracked videos turned out awesome!

1 Like

@bgolson09 You are welcome :slight_smile: May I ask which kind of information you are extracting from these trajectories?
Maybe it is interesting to add them to TraJ ?