Empirical scatter correction: CBCT scatter artifact reduction without prior information.

Area of Science:

• Medical Imaging
• Computational Imaging
• Radiological Physics

Background:

• Cone beam CT (CBCT) image quality is degraded by scattered radiation, causing artifacts like cupping and streak artifacts.
• These artifacts reduce image contrast and distort CT values, necessitating correction methods.
• Existing software solutions often require specific system or scan parameter knowledge.

Purpose of the Study:

• To propose a simple, effective postprocessing software-based method for correcting scatter artifacts in CBCT scans.
• To develop a correction technique that does not rely on prior knowledge of the CT system or scan parameters.

Main Methods:

• Empirical Scatter Correction (ESC) generates scatter-like basis images via convolution of projection data.
• A linear combination of basis images is subtracted from the original projection, followed by logarithm and FDK reconstruction.
• Coefficients for the linear combination are automatically determined using a downhill simplex algorithm to minimize scatter artifacts.

Main Results:

• ESC significantly improved CBCT image quality in simulations and phantom measurements, reducing CT value errors.
• For simulated head CT, ESC reduced deviation from scatter-free reference to -6 HU (uncorrected: > -200 HU).
• ESC enhanced visibility and segmentability of anatomical features in thorax and abdomen scans and improved soft tissue CT values in phantom scans.

Conclusions:

• Empirical Scatter Correction (ESC) effectively reduces scatter artifacts in CBCT using only projection data.
• The method demonstrates comparable results to existing specialized correction algorithms.
• ESC offers a valuable tool for enhancing CBCT image quality without requiring extensive prior information.