jcouto / wfield

Tools to process widefield data
GNU General Public License v3.0
13 stars 14 forks source link

Questions about LEDs and hemodynamics correction #4

Closed oraby closed 4 months ago

oraby commented 3 years ago

Hello,

It was recommended in the paper to set the LED power between 10–25 mW for both LEDs. My question is:

We are still adapting the pipeline to our data, and currently we are having the following reconstructed output of the averaged SVT signal of the reward epoch: Reward_avgVideo The reward is delivered around frame no. ~3-4 , and it's bit unexpected to see the brain gets inhibited as such. We were wondering whether one of the LEDs power is less than what's needed. We currently calibrate our LEDs in continuous mode. Have you experienced such case before?

oraby commented 3 years ago

This was the hemodynamics correction results that we got for this session: image

In this session, I was using the baseline as the 20th percentile of each channel. Does the violet scaling here looks like in order or is it too much?

jcouto commented 3 years ago

Hi Hatem,

that is correct. We are successful with 10-25mW measured in continuous mode (e.g. with a S425C sensor that is large enough to capture all light at the focal plane). This will be an over-estimate of the actual value that reaches the brain but it simplifies the calibration since it does not depend on the imaging rate.

When imaging we first saturate the camera sensor, and then reduce the power so that there are no pixels in cortex with values over 75% of the dynamic range of the sensor (we do this separately for the 470 and 405nm channels).

I don't often see negative regression coefficients, could you run the pipeline with wfield preprocess <foldername> please? I wonder if the baseline could be causing this of if it is something else. Please make sure you do not have a 'frames_average' file in foldername.

Thanks in advance, Joao