A practical cone-beam CT scatter correction method with optimized Monte Carlo simulations for image-guided radiation...

Area of Science:

• Medical Physics
• Radiological Imaging
• Computational Imaging

Background:

• Cone-beam CT (CBCT) is crucial for image-guided radiation therapy (IGRT).
• Scattered photons in CBCT degrade image quality, necessitating correction methods.
• Kernel-based methods are routine, but Monte Carlo (MC) simulations offer higher accuracy despite computational challenges.

Purpose of the Study:

• To develop a practical, efficient, and accurate MC-based scatter estimation and removal method for CBCT.
• To enable routine clinical application of MC-based scatter correction in IGRT.

Main Methods:

• Developed an MC scatter estimation method using planning CT images, not scatter-contaminated CBCT images.
• Implemented a workflow involving rigid registration, MC simulation for scatter estimation, and scatter removal from projections before CBCT reconstruction.
• Optimized the workflow on a GPU platform, incorporating projection denoising, CT downsampling, and angular interpolation for speed.

Main Results:

• Reduced mean HU errors in simulation from 44 to 3 HU (full-fan) and 78 to 9 HU (half-fan).
• Significantly reduced scatter-induced artifacts, such as ring artifacts, in phantom and patient cases.
• Achieved a total computation time of under 30 seconds for scatter estimation and CBCT reconstruction.

Conclusions:

• The developed MC-based scatter correction method is accurate and efficient for CBCT.
• The method's speed and effectiveness make it a promising tool for clinical IGRT applications.
• This approach overcomes the traditional computational burden of MC simulations for scatter correction.