CNMF Errors/Trouble Shooting

[ViciousPlants]

My setup:

1. MacOS
2. M1Pro and RAM: 16 gb
3. Python Version: 3.11.11
4. Caiman version: 1.11.3
5. Which demo exhibits the problem: demo_pipeline.ipynb
6. How you installed Caiman: miniforge
7. Details: Issues with identifying the correct parameters for CNMF for data set.

Hello! I have recently been trying to use Caiman to determine calcium activity in response to a stimulus. Our group has no experience analyzing calcium data before so I have just been following the demos and applying our data to the demos.

When I ran CNMF on my data set (2D timelapse image with 31 time frames), I encountered the following error:

I am not sure what the "nperseg" variable represent and if it is critical.

The components found from CNMF were all incredibly small:

However, I did try to modify the critical CNMF parameters:

Please let me your thoughts and advice on things to try for correctly identifying components.

More information on my dataset:
My raw data is 3D with 103 timepoints (each timepoint is 40 seconds). The test data I am using to learn Caiman is a z-slice of my 3D data.

[kushalkolar]

Hello! I have recently been trying to use Caiman to determine calcium activity in response to a stimulus. Our group has no experience analyzing calcium data before so I have just been following the demos and applying our data to the demos.

Andrea's videos are the best place to start if you haven't watched already:

https://github.com/flatironinstitute/CaImAn?tab=readme-ov-file#videos

When I ran CNMF on my data set (2D timelapse image with 31 time frames), I encountered the following error: I am not sure what the "nperseg" variable represent and if it is critical.

This is a warning not an error

The components found from CNMF were all incredibly small:

I can't say much from this, what kinds of cells are these, are they homogenous in size, what gSig did you use. You can post a video in a github comment (just drag & drop) if it's under 10MB.

More information on my dataset: My raw data is 3D with 103 timepoints (each timepoint is 40 seconds). The test data I am using to learn Caiman is a z-slice of my 3D data.

Are you only analyzing ~100 frames of data? What exactly is this, it doesn't seem like neuronal imaging which is what caiman is meant for. At the least you must not use deconvolution because it makes no sense if your imaging rate is 40 seconds per frame.