search

royerlab / ultrack

Cell tracking and segmentation software
https://royerlab.github.io/ultrack
BSD 3-Clause "New" or "Revised" License
95 stars 13 forks source link

Application to 3D images of mouse brain #120

Closed GiangTranHU closed 4 months ago

GiangTranHU commented 4 months ago

I am new to use ultrack, may I ask a question about it? May ultrack be used to detect cells in 3D images of mouse brain? Thank you very much!

JoOkuma commented 4 months ago

Hey @GiangTranHU, is it live imaging or fixed tissue?

Our main use case is for live imaging. If it's fixed tissue, a subset of image processing functions could be useful; see them here. Especially detect_foreground, which is very good at detecting cells, you can run watershed from skimage.

GiangTranHU commented 4 months ago

Hi JoOkuma, Thank you very much for your help! I work with fixed tissue, I will try with detect_foreground from your program. I also see you applied SAM in your program, may you please give me advice if SAM is useful for my case, especially in 3D images of cells?

JoOkuma commented 4 months ago

@GiangTranHU , SAM didn't work that well for us. We mainly use cellpose in-house. It might also be worth looking at microSAM.

GiangTranHU commented 4 months ago

Thank JoOkuma for help! May I ask a question about cellpose? Does cellpose need to re-train the model before using?

JoOkuma commented 4 months ago

@GiangTranHU, most of the time, we don't need to re-train for 2D images. For 3D, we usually fine-tune (re-train) our models, but then you need to annotate some data :(.

Others might have had a different experience. The imageSC forum could provide additional info.

GiangTranHU commented 4 months ago

Thank JoOkuma!