Fast eikonal phase retrieval for high-throughput beamlines

Area of Science:

• Computational imaging
• Wave optics
• Phase contrast microscopy

Background:

• Phase retrieval is crucial for reconstructing images from wave intensity measurements.
• Existing eikonal phase retrieval methods can be computationally intensive, limiting their application in high-throughput experiments.
• Accurate phase retrieval is essential for quantitative phase-contrast imaging and tomography.

Purpose of the Study:

• To develop a computationally efficient and accurate eikonal phase retrieval formulation.
• To accelerate phase retrieval by over two orders of magnitude while maintaining controlled accuracy.
• To enable high-throughput propagation-based phase-contrast micro-tomography.

Main Methods:

• Developed a fast eikonal phase retrieval method based on a second-order asymptotic expansion and Wentzel-Kramers-Brillouin (WKB) correction.
• Employed an iterative correction scheme with fast Fourier transform (FFT)-diagonal, energy-dependent inverse operators (Paganin-type filters).
• Combined local and non-local solvers for sub-pixel and multi-pixel shifts, utilizing explicit eikonal ray mapping and adjoint operators for robustness.

Main Results:

• Achieved acceleration of eikonal phase retrieval by more than two orders of magnitude.
• Demonstrated controlled accuracy across various experimental regimes, including sub-pixel and multi-pixel shifts.
• Successfully integrated polychromatic data support and GPU/FFT efficiency for practical applications.

Conclusions:

• The introduced fast eikonal phase retrieval formulation offers significant computational speed-up and accuracy.
• This unified approach is suitable for high-throughput propagation-based phase-contrast micro-tomography.
• The method enhances the efficiency and applicability of phase retrieval in advanced imaging.