MouseLand / suite2p

cell detection in calcium imaging recordings
http://www.suite2p.org
GNU General Public License v3.0
334 stars 239 forks source link

Manual labeling issues #1076

Open Kojot1444 opened 6 months ago

Kojot1444 commented 6 months ago

Describe the issue:

Hi, I am using the latest version of suite2P and getting some issues with the manual labeling option. In the manual labeling interface the image appears in a smaller size so that the cells on the edges are not visible and can not be labeled. I am attaching the screenshots below. An other question, I would like to use manual labeling option to track the same cells collected at different days. Is there a way to save the Ids of the marked cells? So that the cell that I mark first would have Id1 and so on?

Capture Capture_manual labling

Thanks a lot!

Reproduce the code example:

I do not get any errors in the code

Error message:

No response

Version information:

v0.14.0

Context for the issue:

No response

chriski777 commented 6 months ago

Hi @Kojot1444, for the manual labeling issue, is this a 1P recording? If so, you can try playing with the spatial_taper parameter as it determines the number of pixels that are ignored for analysis on the edges. You can try lowering it a bit to see if you can get the pixels involved in the cells you want to manually add. There is a tradeoff though between including more pixels in the edges and the quality of the registration so keep that in mind!

As for your second point on being able to have labels across multiple days, we don't explicitly yet have features to accommodate that. One thing you can do is just run the datasets for multiple days together through suite2p. Then, your labels will be consistent across days. This assumes though that the FOV is roughly similar across days.

Kojot1444 commented 6 months ago

Hi, thank you for reply! I am using 2P and gCamp8m. Is there any other parameters for 2P that I could change for analysis on the edges? Thanks!