PIPN: Physics-inspired phase retrieval network for propagation-based X-ray phase-contrast imaging.

Area of Science:

• Medical Imaging
• Computational Physics
• Biomedical Engineering

Background:

• Propagation-based X-ray phase-contrast imaging (PB-XPCI) offers high-resolution soft tissue visualization.
• Phase retrieval from single-distance intensity measurements is an underdetermined inverse problem, often requiring approximations.
• Existing methods like deep learning (DL) demand extensive training data or complex computations.

Purpose of the Study:

• To develop a data-free, physics-based phase retrieval method for PB-XPCI.
• To introduce an accelerated physics-inspired phase retrieval network (PIPN).
• To overcome limitations of conventional and DL-based phase retrieval techniques.

Main Methods:

• A novel physics-inspired phase retrieval network (PIPN) was developed.
• The PIPN utilizes a single approximation condition and a physics imaging model.
• An acceleration strategy was implemented for the PIPN.

Main Results:

• PIPN achieved phase retrieval without requiring training data.
• The accelerated PIPN rapidly reconstructed high-quality phase projections.
• The method demonstrated stability across various propagation distances.

Conclusions:

• The proposed PIPN offers an efficient and data-free solution for X-ray phase retrieval.
• The acceleration strategy enhances reconstruction speed and maintains image quality.
• PIPN advances PB-XPCI for soft tissue imaging applications.