Open alishbaimran opened 2 days ago
@Soorya19Pradeep looks like the fluorescence channel in registerd.zarr has zeros (this can be due to noise) and the phase channel has few saturated pixels. Please look into this and figure out why we are getting clipped pixel values in the registration process.
@alishbaimran this does not block the current iteration on model and dataloader.
Summary and expected behavior
I'm getting a white square (values seem to be 1 when I hover over the square in napari) showing up in napari when I try to visualize data for some fovs and specific cells/patches. It seems that the bug is coming from my registered.zarr. The path to the registered file is:
/hpc/projects/intracellular_dashboard/viral-sensor/2024_02_04_A549_DENV_ZIKV_timelapse/2-register/registered.zarr
.The specific cells I see the issue in is A/3/0 6th timestep and 36th z slice. channel 1 is rfp and channel 3 is phase. These are the channels we are looking at.
The patches data can be found at:
/hpc/projects/intracellular_dashboard/viral-sensor/2024_02_04_A549_DENV_ZIKV_timelapse/6-patches/patch_final_2.zarr
. The specific cells I am getting the issue with are at/hpc/projects/intracellular_dashboard/viral-sensor/2024_02_04_A549_DENV_ZIKV_timelapse/6-patches/patch_final_2.zarr/A/3/fov0cell10
and/hpc/projects/intracellular_dashboard/viral-sensor/2024_02_04_A549_DENV_ZIKV_timelapse/6-patches/patch_final_2.zarr/A/3/fov0cell11
around timestep 5/6th and 36th z slice.I've plotted the following histograms from the patches ome_zarr for A/3/fov0cell11 at timestep 6, z slice 36. Channel 0 is rfp and channel 1 on the diagram is phase.
![image](https://github.com/mehta-lab/recOrder/assets/44557946/a1c4bed9-31a4-4431-bc65-944f3fe253e9)
This is the registered file data from A/3/0 at timestep 6, z slice 36. Channel 1 is rfp and channel 3 on the diagram is phase.
![image](https://github.com/mehta-lab/recOrder/assets/44557946/4051623a-f7da-4b64-b25d-349640454842)