Deduplicating stage interpolation methods

[keithchev]

Overview

"Stage interpolation" refers to the process by which the z-position of the focal planes for an array of samples is interpolated from the measured z-positions of a subset of those samples. This process is necessary when imaging on large coverslips (e.g., a 96-well plate) because the z-position of the sample focal plane varies across the coverslip due to its nonzero tilt and curvature, and it is not practical to measure the z-position for each sample. For example, in the case of a 96-well plate, one simple approach would be to measure the z-position of the focal plane in only the four corner wells, then interpolate the z-position of all the other wells.

Because we often image on 96-well plates at the Biohub, this problem comes up a lot, and it may make sense to de-duplicate these efforts.

Current solutions

• opencell microscopy automation (the dragonfly-automation repo): this micromanager-based project uses an interactive jupyter notebook to allow the user to visit a subset of wells on the plate and measure the z-position of the focal plane in those wells. Then a simple interpolation method is used to interpolate the z-positions for the unvisited wells. Finally, the interpolated positions are added to the micromanager position list. Note that the complicated part of this process is the code/logic needed to support the interactive jupyter notebook, not the interpolation itself.

Possible solutions

Once we have an understanding of the current approaches to this problem, we can consider whether and how to merge them or implement a new, general-purpose solution.

[keithchev]

@edyoshikun, @i-jey, and @YangLiujames can you comment here (or edit the original issue) to describe how you are currently solving (or thinking about solving) this problem? Also please tag anyone else who would is working on this!