poly-DART: A discrete algebraic reconstruction technique for polychromatic X-ray CT

Area of Science:

• Medical Imaging
• Computational Imaging
• Materials Science

Background:

• Discrete Algebraic Reconstruction Technique (DART) excels in monochromatic X-ray CT, leveraging material number knowledge for high-quality reconstructions.
• Polychromatic X-ray sources, common in practical applications, introduce beam hardening artifacts that significantly impair DART's performance.
• Existing DART methods struggle with the complexities of polychromatic X-ray data, limiting their applicability.

Purpose of the Study:

• To introduce poly-DART, a novel discrete tomography algorithm designed for polychromatic X-ray sources.
• To address the limitations of DART caused by beam hardening effects in polychromatic X-ray CT.
• To improve image segmentation accuracy using sparse attenuation information within a DART framework.

Main Methods:

• Developed the poly-DART algorithm, integrating sparsity exploitation with polychromatic data handling.
• Employed DART principles to reconstruct images while accounting for the spectral characteristics of X-ray sources.
• Validated the algorithm using Monte Carlo simulations and experimental polychromatic X-ray data.

Main Results:

• Poly-DART demonstrated significantly improved image segmentation compared to standard DART on polychromatic data.
• The algorithm effectively mitigated beam hardening artifacts, leading to more accurate reconstructions.
• Both simulated and experimental results confirmed the superior performance of poly-DART.

Conclusions:

• Poly-DART offers a robust solution for discrete tomography with polychromatic X-ray sources.
• The algorithm enhances image quality and segmentation accuracy in challenging X-ray CT scenarios.
• This advancement broadens the applicability of DART-based methods in real-world X-ray imaging.