Advanced Phase-Retrieval for Stepping-Free X-Ray Dark-Field Computed Tomography

Area of Science:

• Medical Imaging
• X-ray Technology
• Computational Imaging

Background:

• Grating-based X-ray imaging provides phase-contrast and dark-field channels for enhanced soft tissue and microstructural information.
• Conventional methods face challenges in minimizing data acquisition time for medical applications, particularly in dark-field computed tomography.
• Continuous detector movement, ideal for speed, conflicts with traditional phase-stepping techniques.

Purpose of the Study:

• To address the challenge of crosstalk artifacts in grating-based dark-field CT with continuous data acquisition.
• To introduce a novel demodulation-based interpretation of phase retrieval for continuous acquisitions.
• To present algorithmic extensions for improved phase retrieval and crosstalk mitigation.

Main Methods:

• Reinterpreting phase retrieval as a demodulation problem to understand crosstalk origins.
• Developing three algorithmic extensions to the conventional sliding-window approach.
• Validating algorithms through simulation studies and experimental data from a human-scale dark-field CT prototype.

Main Results:

• Identified partially overlapping modulation side bands as the source of crosstalk artifacts.
• Demonstrated substantial reduction in crosstalk artifacts using the proposed algorithms.
• Confirmed algorithm effectiveness in both simulated and experimental scenarios.

Conclusions:

• The novel demodulation-based approach effectively explains and mitigates crosstalk artifacts in continuous acquisition grating-based dark-field CT.
• The presented algorithmic extensions significantly improve phase retrieval accuracy and image quality.
• These advancements pave the way for faster and more reliable medical applications of dark-field CT.