search

AllenNeuralDynamics / aind-fip-dff

This capsule takes an input raw NWB file of fiber data and creates processed DFF traces
MIT License
0 stars 0 forks source link

Evaluation of FIP preprocessing #12

Open alexpiet opened 2 months ago

alexpiet commented 2 months ago

@hagikent Can you comment on each of these examples. Are they problems? Do they have known causes or solutions?

alexpiet commented 2 months ago

Session with poor bleaching subtraction:

@hagikent thinks that G1 could have poor signal. R channel has no sensor.

behavior_717531_2024-08-23_08-58-49 Image

alexpiet commented 2 months ago

Flat signal at the start of the session. Is this because excitation wasn't running? Or is this an artifact of processing?

@hagikent says this is intentional. and we have a blank control to estimate floor of sensor @hagikent says that start signal should be exactly flat. behavior_711042_2024-08-23_13-27-08

Image

alexpiet commented 2 months ago

Session with G_1_preprocessed and G_2_preprocessed all NAN. G_1 and G_2 are NOT all NaN behavior_708031_2024-08-26_08-49-03

alexpiet commented 2 months ago

Strange artifacts in FIP signal behavior_717531_2024-08-26_09-08-42

@j-friedrich thinks this could be an artifact of converting between nwb and pandas dataframe

Image

jeromelecoq commented 1 month ago

this need @hagikent eyes. @j-friedrich made PR and applied to real and synthetic data.

jeromelecoq commented 3 weeks ago

@hagikent will use the repo from @rachelstephlee to check sessions and start checking these.

alexpiet commented 3 weeks ago

repo that contains notebook for looking at data: https://github.com/AllenNeuralDynamics/aind_ssm_hackathon

hagikent commented 3 weeks ago

wait for motion correction integration

hagikent commented 3 days ago

@j-friedrich I still see you are pushing commits to the MC PR. Still working on it or should I start testing it?