royerlab / ultrack

Cell tracking and segmentation software
https://royerlab.github.io/ultrack
BSD 3-Clause "New" or "Revised" License
64 stars 7 forks source link

tracking parameters related to the weighting of the raw intensity values #95

Closed gouthamr321 closed 3 months ago

gouthamr321 commented 3 months ago

Hello,

Very cool work!

When running the tracking (after we already have segmentations), I see that we are passing in the raw images themselves to the track function in ultrack/core/main.py. I understand that that is because the florescence intensity is used for the linking steps. But for cases where we are tracking nuclei from brightfield, where the raw images don't have as meaningful intensity gradients, Is there a way to change the weight of the importance of the raw intensity values that are extracted for the ILP? Some specific parameter in the tracking_config schema perhaps?

Thanks!

JoOkuma commented 3 months ago

Hey @gouthamr321,

We only use fluorescence information for tracking if you provide the images parameter of the track function. Otherwise, we use the cells' segments mask to compute the association between segments.

def track(
    config: MainConfig,
    *,
    labels,
    sigma,
    detection,
    edges,
    images,   # <- don't use this in your case.
    scale,
    vector_field,
    overwrite,
):

Feel free to drop other questions :)

gouthamr321 commented 3 months ago

Thanks for the clarification! Just confirming that the tracking looks good in my case when not passing in the raw images!

JoOkuma commented 3 months ago

Glad to hear @gouthamr321! I'll close the issue.