motion correction failures on Mesoscope sessions

AllenInstitute / ophys_etl_pipelines

Pipelines and modules for processing optical physiology data

Other

9 stars 5 forks source link

motion correction failures on Mesoscope sessions #203

Open djkapner opened 3 years ago

djkapner commented 3 years ago

from @nataliaorlova :

Hi Dan and Wayne, I noticed this weird artifact (squares around the cell bodies) in one of the planes on a test Meso2 session (1090674621), looked at the video and it seems to me that motion correction is the cause. Can you please take a look at the mc preview video and let me know your thoughts, should we try and re-run the mc for this plane?

For that session (8 experiments should share the same motion) 6 had decent motion correction and 2 did not:

@nataliaorlova reports that those planes don't seem to have any apparent problem from imaging quality.

Tasks:

[ ] try to find some reason why these aren't working well
[ ] could try Suite2P double-pass from this branch NOTE: this did not help deepscope planes that had low SNR.
[ ] could try the latest version of Suite2P (v0.10.0) Some info in the slides indicates they have "fixed some registration issues"

Validation:

[ ] Either a data-reason why these fail (i.e. SNR very low) or a strategy for registering.

samiamseid commented 3 years ago

I have two more experiment planes that have motion correction issues.

The first one looks like its just correcting to the wrong cells at the beginning "\allen\programs\braintv\production\visualbehavior\prod5\specimen_1075943220\ophys_session_1100665830\ophys_experiment_1100880610\processed\1100880610_suite2p_motion_preview.webm"

This other has the distinctive square pattern "\allen\programs\braintv\production\visualbehavior\prod5\specimen_1075943220\ophys_session_1100665830\ophys_experiment_1100880616\processed\1100880616_suite2p_motion_preview.webm"

samiamseid commented 3 years ago

I have a session where the output of the motion correction seems to be shifted a baseline of around 90 to 100 pixels. None of the other planes are shifted this much, and the SNR of this plane seems fine to me.

Experiment 1109580767

"\allen\programs\braintv\production\visualbehavior\prod6\specimen_1091244055\ophys_session_1109456332\ophys_experiment_1109580767\processed\1109580767_suite2p_motion_preview.webm"

motcor

danielsf commented 3 years ago

@samiamseid We were just "admiring" that failure. We think something might be wrong with the way the TIFF files coming off Mesoscope are being split. This issue

https://github.com/AllenInstitute/ophys_etl_pipelines/issues/257

is where we are intending to track that work

nataliaorlova commented 3 years ago

Hi all, we have another session where we see signal in pre-motion correction data, and to my eye the signal doesn't look particularly low given that this is an inhibitory line. But after motion correction he signal is gone.

here is an example MC preview for one of the planes, but all 8 have similar issue: \allen\programs\braintv\production\visualbehavior\prod5\specimen_1078949119\ophys_session_1109681854\ophys_experiment_1109780813\processed\1109780813_suite2p_motion_preview.webm

Session ID is 1109681854

Logs seems to have some warning out offsets being clipped, sounds like the MC might have failed on this session. Do we want to try and re-run it?

Also, forgive me if this should be a separate issue... there are several related to the failures of MC on Meso data and I'm a bit confused where this should be reported.

danielsf commented 3 years ago

For the record, all of the experiments in that ophys session failed CELL_ROI_CREATION with no ROIs found. I just backchanneled Natalia to discuss this.