PSII photo processing: open questions for discussion

[ghost]

• Are we saturating the PSII? does it matter for comparison across genotypes?
  • LEDs are (much) more powerful than Danforth
  • can we experimentally determine saturating light?
• Can we map 3D structure (distance + angle) to the PSII output ( @solmazhajmohammadi ?)
  • can a 3d point cloud be used to identify leaves and distance of leaves from imager?
  • and can we use this to correct?
  • how does F relate to distance (See inverse square law F ~ (1/distance^2) this should be applied to fluorescence from leaf to sensor; can we ignore it from LEDs to leaf (if it is still saturating? otherwise we may need to measure / empirically derive the nonlinear function.
• need to segment leaves

How to plot data?

Lemnatec sent test data to Noah to compare hand measurements with camera data. They put the leaves parallel to the ground / sensor, which will not be relevant for field measurements.

https://github.com/terraref/extractors-multispectral/tree/master/psii_fluorescence

TODO:

• process test data from Lemnatec to determine if the camera values are relevant. ( @nfahlgren )
• evaluate segmenting ( @ZongyangLi ) and doing a rough calculation of Fv/Fm
  • Fmax and Fmean look the same
  • frame 1=fdark, frame2=fmin, frame3=fmax. Only use these frames.
• decide how to evaluate / correct for whether illumination is saturating
  • @smarshall-bmr have you run tests with lights at different intensity?

Completion criteria?

• protocol for using PSII camera
• convert protocol into next steps toward generating Fv/Fm as a phenotype
  • Currently covered by Write extractor for PSII camera #152; that issue may need to be split up into separate issues.

[smarshall-bmr]

While giving tours we had a specialist in Photosystem II express doubt that we would be able to saturate the system @TinoDornbusch knows more about this then I do though. Theoretically we should see saturation as "clipping" on the response curve.

I have taken data sets at varying light intensities, exposure times, different distances apart (to characterize leakover distances) and different times apart (to asses dark adaption times.)

Much of this will change however, I'm planning on modifying the focal distance of the camera in order to give us a bit larger margin of error when it comes to scanning the field, currently the scanner box must be within 30cm of the plants to take a measurement, too close for me to comfortably allow the system to run autonomously.

[dlebauer]

@smarshall-bmr the question of how to address this criticism should be captured in the first bullet point in the issue 'Are we saturating the PSII? does it matter for comparison across genotypes?'

What do you mean by 'clipping'? Can you link (or give the dates of) the datasets that you generated at varying light intensities and propose a method of testing if we are saturating PSII?

[dlebauer]

@ZongyangLi and @nfahlgren can you please summarize the results of your meeting, update this isssue, and create new ones as necessary?

[ZongyangLi]

From discussion with @nfahlgren , I am planning to create 2 extractors for this:

• convert bin files into gray scale images,

• create pseudocolored Fv/Fm image using F-images and create a Fv/Fm histogram.

[smarshall-bmr]

@ZongyangLi @nfahlgren LemnaTec prepared an Octave script for me to use in order to check data that produces a Fv/Fm image as well as a response histogram. Would this be helpful for you?

[ZongyangLi]

@smarshall-bmr Thanks a lot! Is this the LemnaTec script you mentioned:
https://github.com/terraref/computing-pipeline/blob/master/scripts/PS2/GetPS2.m

I think I am writing script like this, except the way of determining a Fmax.

[TinoDornbusch]

@nfahlgren @ZongyangLi I am in the Loop now and happy to contribute to write the Extractor. Can anyone help me to get Clowder running and briefly Show me how it works?

I would then share this Knowledge with my colleagues.

[nfahlgren]

@TinoDornbusch: @ZongyangLi and @max-zilla created a version 1 extractor for PSII that converts the raw images to PNG and generates a histogram of Fv/Fm on the whole field-of-view. The code is here: https://github.com/terraref/extractors-multispectral/tree/master/psii_fluorescence. Right now it doesn't subset the plant pixels or generate any traits (e.g. mean Fv/Fm, etc.) but the plan was to extend the extractor after getting the basic infrastructure in place. We are happy to work with you on it.

[TinoDornbusch]

@nfahlgren @ZongyangLi @max-zilla Thanks, I will be in Maricopa during the next two weeks and plan to work on the PSII analysis. lets discuss some "trial experiments". I had treatment with PS2 blocking herbicide in mind and to monitor the PS2 efficiency over time.

[TinoDornbusch]

@smarshall-bmr Stuart you think it might be possible to get some herbicide containing DCMU. In Europe this is not available anymore.

https://en.wikipedia.org/wiki/DCMU

[dlebauer]

https://terraref.ncsa.illinois.edu/clowder/datasets/5936a3ad4f0c0ead2da27389 example output