Separable scatter model of the detector and object contributions using continuously thickness-adapted kernels in CBCT

Area of Science:

• Medical Physics
• Image Reconstruction
• Non-Destructive Testing

Background:

• Cone Beam Computed Tomography (CBCT) systems utilize larger X-ray illumination fields, increasing scatter radiation.
• Existing reconstruction algorithms often neglect scatter radiation, leading to artifacts in reconstructed images.
• High-energy X-ray sources (hundreds of keV) in industrial Non-Destructive Testing (NDT) amplify detector scatter, posing reconstruction challenges.

Purpose of the Study:

• To comprehensively evaluate the contribution of detector scatter to artifacts in CBCT imaging.
• To develop and present a method for separating scatter originating from the object versus the detector.
• To demonstrate the insufficiency of object-only scatter correction for artifact removal.

Main Methods:

• Utilized continuously thickness-adapted kernels for evaluating detector scatter.
• Employed a four-Gaussian model to differentiate between object scatter and detector scatter.
• Performed experimental validation using a collimator to isolate object scatter effects.

Main Results:

• Detector scatter significantly contributes to artifacts in CBCT reconstructions, comparable to object scatter.
• Scatter correction solely addressing the object is insufficient for artifact-free imaging.
• Experimental validation confirmed the substantial impact of detector scatter.

Conclusions:

• Accurate scatter correction in CBCT, especially for NDT applications, must account for both object and detector scatter.
• The proposed four-Gaussian model effectively separates scatter sources.
• Improved image quality in high-energy CBCT requires comprehensive scatter mitigation strategies.