Iterative correction of beam hardening artifacts in CT

Area of Science:

• Medical Imaging
• Computed Tomography (CT)
• Image Reconstruction

Background:

• Beam hardening artifacts are a significant challenge in CT imaging, arising from the polychromatic nature of X-ray beams.
• These artifacts can distort image data, particularly in regions with varying material compositions, impacting diagnostic accuracy.
• Existing correction methods often require prior knowledge of the X-ray spectrum and object materials, limiting their applicability.

Purpose of the Study:

• To develop and evaluate novel methods for reducing beam hardening artifacts in CT.
• To address the challenge of artifact reduction when X-ray spectrum and material properties are unknown.
• To improve the homogeneity of reconstructed CT images, particularly in uniform regions.

Main Methods:

• The study assumes objects can be segmented into materials with distinct attenuation coefficients and parameterizes the X-ray spectrum using energy bins.
• Unknown spectrum parameters and material attenuation values are estimated by minimizing the difference between measured and simulated polychromatic sinogram data.
• Three iterative algorithms were derived: two for image reconstruction (IGR, IFR) and one for sinogram pre-correction (ISP).

Main Results:

• All three developed methods (IGR, IFR, ISP) successfully reduced cupping artifacts in real X-ray data from high and low-contrast phantoms.
• Reconstructions achieved accurate homogeneity in segmented regions, even with imperfect initial image segmentation.
• The methods demonstrated robustness, tolerating significant variations in material uniformity within segments.

Conclusions:

• Effective beam hardening correction in CT is achievable without prior knowledge of material properties or the X-ray spectrum.
• The proposed iterative algorithms offer a viable solution for artifact reduction in diverse clinical scenarios.
• These findings enhance the reliability and accuracy of CT imaging for material characterization and diagnosis.