Non-linear regularized phase retrieval for unidirectional X-ray differential phase contrast radiography.

Area of Science:

• Medical Imaging
• Computational Physics
• Image Processing

Background:

• Phase retrieval from unidirectional radiographic differential phase contrast (DPC) images is challenging due to noise and artifacts.
• Direct integration of noisy DPC data often leads to stripe artifacts, degrading image quality and retrieval accuracy.

Purpose of the Study:

• To develop and validate a purely algorithmic method for suppressing stripe artifacts in DPC phase retrieval.
• To improve the accuracy and contrast of phase images obtained from unidirectional DPC data.
• To provide a versatile tool for 2D imaging applications utilizing differential data.

Main Methods:

• A novel algorithmic approach is presented to mitigate stripe artifacts during the phase retrieval process.
• The method relies on mathematical processing, avoiding the need for additional hardware alignment or extended scanning times.
• Numerical simulations and experimental data were used to evaluate the method's performance.

Main Results:

• The proposed method effectively suppresses stripe artifacts commonly observed in direct integration techniques.
• Significant improvements in phase retrieval accuracy were demonstrated.
• Enhanced contrast in the resulting phase images was observed, leading to clearer visualization.

Conclusions:

• The developed algorithmic method offers a robust solution for artifact reduction in DPC phase retrieval.
• Its independence from hardware modifications makes it a practical and widely applicable tool.
• The technique enhances the utility of DPC imaging in various 2D applications requiring precise phase information.