Segmentation-free empirical beam hardening correction for CT

Area of Science:

• Medical Imaging
• Computational Imaging
• Image Reconstruction

Background:

• Standard CT image reconstruction often overlooks polychromatic X-ray beam effects, causing cupping and beam hardening artifacts.
• Existing correction methods like water precorrection or physical/empirical approaches often require tissue segmentation, which can be inaccurate due to artifacts themselves.

Purpose of the Study:

• To introduce a novel segmentation-free empirical beam hardening correction (sfEBHC) algorithm.
• To address limitations of existing methods by eliminating the need for tissue segmentation, improving artifact correction in noisy and artifact-degraded images.

Main Methods:

• A histogram deformation technique is applied to the primary CT image, creating a segmentation-free (sf) approach.
• Monochromatic forward projections of the original and deformed volumes are combined and reconstructed into basis volumes.
• Image flatness is maximized by adding a weighted sum of basis images for artifact correction.

Main Results:

• The sfEBHC method demonstrates equal or improved image quality compared to standard empirical beam hardening correction (EBHC).
• Effective correction of beam hardening artifacts across various scan parameters, scenarios, and imaging devices (dual source spiral CT, digital volume tomograph, micro CT).

Conclusions:

• sfEBHC successfully reduces beam hardening artifacts, enhancing image quality in challenging datasets with high noise and severe artifacts.
• The algorithm can recover structures that are obscured in beam hardening-affected regions, improving diagnostic utility.