Phase and absorbance retrieval in X-ray holographic microscopy under weak illumination using physics-driven neural...

Area of Science:

• Physics
• Materials Science
• Biophysics

Background:

• X-ray holographic microscopy offers nanoscale resolution for 3D imaging of morphology and phase contrast.
• Recovering phase and absorbance from shot-noise-limited (SNL) X-ray holograms under weak illumination presents significant challenges.

Purpose of the Study:

• To develop a deep learning model for accurate single-shot phase and absorbance retrieval from SNL X-ray holograms.
• To reconstruct 3D complex wavefield, phase, and absorbance distributions directly from limited data.

Main Methods:

• A physics-driven neural network, MorpHoloNet-X, was developed.
• The model incorporates physics-based prior knowledge and wave propagation principles.
• Performance was validated using synthetic and experimental SNL holograms.

Main Results:

• MorpHoloNet-X successfully reconstructed 3D complex wavefield, phase, and absorbance.
• The model demonstrated effective phase and absorbance retrieval from SNL X-ray holograms.
• Results showed comparable or improved performance against conventional methods.

Conclusions:

• The proposed MorpHoloNet-X facilitates robust phase and absorbance retrieval from challenging SNL X-ray holograms.
• This technique is valuable for hard X-ray holography under rapid acquisition or weak illumination conditions.
• MorpHoloNet-X advances nanoscale imaging in biological and material sciences.