Add docs about PMT use/saturation with/w/o shutters

[jmdelahanty]

It was determined recently that the current use of LED stimulations for optogenetic manipulations used in Austin's project is problematic for three main reasons:

• LED light is present in the images obtained during the trials and appear as small banding artifacts throughout the image
• Given above point, there is LED light that is hitting the PMTs. The LED light is much more intense and risks saturating the PMTs and also risks damaging the PMTs/dramatically reducing their effectiveness/wears them down much faster than normal.
• Black bars are present in the images obtained during trials with optogenetic stimulation when the shutter closes.

Each of these things were discussed early on with the optogenetic stimulation implementation for this repository and had the following resolutions at the time (from what I remember):

• Small amounts of data loss during trial periods is not especially problematic if we can interpolate around them or just not use those frames as datapoints
• Since we block at least half of the LED stimulations (we stimulate at 20 Hz while the PMTs close only at 10 Hz), we're reducing PMT exposure as much as we can and that is better than nothing
• See first point

Kay in a recent meeting with me during the week of 8/29/22 told me that Austin should (strongly) consider reducing LED stimulation to 10 Hz to match the PMT shutter speeds so the PMTs are not damaged. However, Austin and I had a recent conversation with Steven Trier on 9/2/22 about this issue and brought up the following:

• A filter is in place already in the imaging path for both our blue and orange LEDs that should block a majority of the light from hitting the PMTs
• A test should be done (ideally with a mouse imaging down a GRIN lens but we don't have any available) with normal imaging first, then with the shutters activated, then with the LED activated, and then finally with both where the values of PMT LUTs are observed and recorded to see if they are saturated.
• This data/results should be sent along to Steve at the associated One Drive link
• Results should be documented and posted to the docs site when possible

This test is being conducted as of writing this issue now! Here's a few items:

• Collect imaging series' with screen recordings of LUTs and images
• Collate data into organized folders on the server
• Send this data to Steve at the one drive
• Add docs to site

[jmdelahanty]

Test was successful and, at least from an initial glance, demonstrated the following (docs forthcoming...):

• Ch1 (used for imaging red fluorophores predominantly) does NOT have a filter and the PMTs are saturated by the LED
• The shutters being closed at the 20Hz rate leaves them closed for quite a while relative to the stimulations obviously (2x the time) and from just staring at the imaging during these tests the number of frames without contamination around them appears to be quite low. Further investigation with Steve and on our own needs to be done to see if we can even reasonably interpolate frames or we need to get Bruker to let us close the shutters much faster (a feature they have not implemented yet)
• Ch2 does have the filter in front of it and does not appear to saturate the PMT
• With the settings Austin uses typically for imaging (Pockels at 550, PMTs about 700-750) the PMTs are saturated around some structures. It would probably be worth it both for image quality and also for PMT health to turn down the PMT high voltage.

On our server in /snlkt/data/Team2P there's a new directory called pmt_testing_led590 which has the data collected from this test. The data has also been uploaded to Bruker's One Drive per Steve's request.

EDIT:

We should consider swapping Ch1 and Ch2's PMTs or purchasing a new PMT to replace it. Ch1 looks much cleaner/has less shot noise, although that could be because the filter in front of it is reducing other light. Worth asking Kevin and Steve about I think.

The Data

2500 images were collected with both PMT channels sampled from which takes approximately 80 seconds at the fastest imaging settings. At 35 seconds, approximately frame 1050, a series of 100 LED stims will occur (5 seconds, 20 Hz) and 50 shutter closings occur (5 seconds, 10 Hz). Each channel has its own folder simply titled ch1 and ch1. Note that the contents of ch2 have images with Ch3 in the filename due to our malfunctioning second channel on our general standards DAC.

Lastly here's a screenshot of one of the images with each artifact in them. The left image is Ch2, our main imaging channel, the right is Ch1, our secondary imaging channel.

Notice that Ch2 (left) has a blue large bar in place (indicating values of 0 as the shutter was closed during those scans) and only a small section with a lighter white bar just above it. This means that a somewhat elevated amount of light is hitting the PMTs, but the filter is certainly helping/blocking a lot of it. Interestingly, the samples themselves are peaking the PMT values! Definitely something we should test some more...

Notice on Ch1 (right) there's a black bar in place (value of 0 as above) and then a bright red bar just above it. This means that the PMTs were getting blasted by the LED and saturating them entirely. We shouldn't test Ch1 any further in this manner unless it's explicitly necessary.

Docs

Docs will be added in the future, but it's going to take some time to thoroughly write out what I did, why, and then what we will choose to do about it.

[jmdelahanty]

To be clear here, the choice is to disable the shutters and just let the PMTs get hit. Note that if/when the lab decides to use dual color imaging, the first channel's PMTs have no filter in place and so either one must be purchased or the shutters would have to be re-enabled.