Phase retrieval from single biomolecule diffraction pattern.

Area of Science:

• Coherent X-ray Diffraction Imaging (CXDI)
• Biomolecular Imaging
• Computational Imaging

Background:

• Conventional phase retrieval methods struggle with low signal-to-noise ratios typical in single biomolecular imaging.
• Femtosecond X-ray Free Electron Laser (XFEL) pulses necessitate advanced imaging techniques for single molecule analysis.
• Existing methods like Hybrid Input-Output (HIO) require object boundary constraints, limiting their applicability.

Purpose of the Study:

• To introduce a novel sparse phase retrieval (SPR) method for Coherent X-ray Diffraction Imaging (CXDI).
• To overcome limitations of conventional methods in noisy, low-signal environments relevant to single biomolecular imaging.
• To develop a phase retrieval technique that does not require explicit object boundary constraints.

Main Methods:

• The proposed SPR method utilizes Bayesian statistics for phase estimation.
• It incorporates a prior distribution defined by an exponential distribution, replacing the need for object boundary constraints.
• The method is evaluated through simulations under noisy conditions with masked central pixels.

Main Results:

• Simulation results demonstrate the SPR method's effectiveness in reconstructing electron density.
• Successful reconstruction was achieved even under significantly noisy measurement conditions.
• The method showed robustness in recovering the object's structure despite the masking of central pixels.

Conclusions:

• The SPR method offers a robust alternative for phase retrieval in CXDI, particularly for challenging datasets.
• Its Bayesian framework and data-driven prior distribution enable accurate reconstruction without object boundary assumptions.
• This advancement holds promise for enabling single biomolecular imaging with femtosecond XFEL sources.