Phase retrieval method for in-line phase contrast x-ray imaging and denoising by regularization.

Area of Science:

• Medical Imaging
• Physics

Background:

• Phase contrast X-ray imaging is gaining popularity due to its advanced capabilities.
• In-line phase contrast X-ray imaging is a promising technique due to its inherent simplicity.

Purpose of the Study:

• To analyze the filtering processes in various non-iterative phase retrieval methods for in-line phase contrast X-ray imaging.
• To address the over-smoothing issue commonly encountered in L(2)-norm regularization methods.

Main Methods:

• Analysis of existing non-iterative phase retrieval methods including Bronnikov, modified Bronnikov, phase-attenuation duality (PAD), single-material, and two-material methods.
• Demonstration that filters in these methods are minimizers of L(2)-norm regularization problems.
• Development of two novel methods to mitigate over-smoothing.

Main Results:

• Established that filters in widely used non-iterative methods are L(2)-norm minimizers.
• Identified the root cause of over-smoothing in these regularization-based approaches.
• Proposed two new methods to effectively overcome the over-smoothing problem.

Conclusions:

• The study provides a unified understanding of filters in non-iterative phase retrieval for in-line phase contrast X-ray imaging.
• The proposed methods offer improved phase retrieval by addressing the limitations of L(2)-norm regularization.