Meniscus correction in phase

[ziw-liu]

Meniscus effect limits the image quality of phase reconstruction at the edge of the well, limiting sample utilization. This problem is most prominent when samples are precious, for example in OPS and iPSC imaging.

Computational correction of the meniscus effect has been reported in other phase modalities:

• Fourier ptychographic microscopy
• QPI

Since our 3D phase transfer function ('s absolute value) has rotational and mirror symmetry along Z, maybe it is possible to directly estimate the angular offset (I'm thinking 2 scalar values for angles in XZ and YX planes) from data, and correct for the effect at reconstruction time.

[ziw-liu]

estimate the angular offset (I'm thinking 2 scalar values for angles in XZ and YX planes) from data

Given the 3D DFT of a non-empty volume without noise, this problem is similar to fitting an ellipse for the sum projection along Z. This sounds like a problem with a least squares solution, i.e. given enough SNR, there should be a unique optimal solution.

Edit: add image

[mattersoflight]

@ziw-liu I love this idea! The orientation of elliptical OTF depends on which side of the NA is clipped, i.e., which side of the wall the illumination is coming from. Perhaps we can model this as an asymmetric Gaussian filter on top of the optical filter that is estimated from spectra.