InCell 1000/2000 field count and plane metadata fixes #3898
Conversation
See https://trello.com/c/gFmUKSp6/362-incell-exposure-times The intended use case is when most planes in the same channel have a consistent exposure time, but a small number have a slightly different exposure time. If a plane-specific exposure time ("PlaneExposureTime_n") is defined in the configuration, it overrides any channel-specific exposure time. Plane-specific times are not written by test-config, but can be added by hand in the small number of cases where they apply. Differences in exposure time between planes in the same channel more commonly indicates a reader bug, so this forces us to check more carefully.
|
This requires discussion, which we're not likely to have time for before 6.12.0. We can reconsider for the next non-patch release. |
There was a problem hiding this comment.
Unless there is a data rationale, my preference would be to avoid injecting artificial blank images especially with HCS data where we should be able to handle sparse data. In addition to the two solutions proposed in the description, a third option could be for the reader to generate a representation as close as possible to the content of the xdce:
Image 0 <-> WellRow 0, WellCol 0, WellSample 0
Image 1 <-> WellRow 0, WellCol 0, WellSample 1
Image 2 <-> WellRow 0, WellCol 1, WellSample 2
Image 3 <-> WellRow 0, WellCol 1, WellSample 3
Image 4 <-> WellRow 0, WellCol 1, WellSample 4
Image 5 <-> WellRow 0, WellCol 2, WellSample 5
Image 6 <-> WellRow 0, WellCol 2, WellSample 6
Image 7 <-> WellRow 0, WellCol 2, WellSample 7
We do have a few examples of sparse plates in the sense that the number populated wells varies as a function of the field index - see for instance https://idr.openmicroscopy.org/webclient/?show=plate-5563. It would be interesting to have a feeling of whether our clients can handle the extreme scenario of this InCell dataset.
If we wanted to make a comparative review of different layouts, would it be useful to try and export the source dataset into a set of various OME-XML/OME-TIFF samples that could be imported?
Fixes #3895 and https://trello.com/c/gFmUKSp6/362-incell-exposure-times
showinf -nopix -omexmlon the .xdce file in QA 31270 should show an exception similar to what is noted in #3895. With this PR, the same command should show 8 series, each of which is a field in well A1.That's almost but not quite correct with respect to what's in the .xdce. The .xdce defines 8 fields, but each field is only present in one of 3 wells (A1, A2, A3). So 8 total series makes sense, but the
Well/WellSamplemapping does not. We would need to decide how to best represent sparse fields here - does 24 series (16 of which are blank) make sense instead? Or is it better to have 8 series and ignore the fact that they are 8 distinct fields?QA 31270 also exposes the exposure time issue noted in https://trello.com/c/gFmUKSp6/362-incell-exposure-times, so the last 2 commits here address that problem as well. This requires some configuration updates, which are in a forthcoming PR.
Opening as draft for now since this requires some discussion around sparse field handling.