[Four-dimensional cone-beam CT reconstruction based on motion-compensated robust principal component analysis].

Area of Science:

• Medical Imaging
• Image Reconstruction
• Computational Imaging

Background:

• Motion artifacts and streak artifacts degrade the quality of cone-beam computed tomography (CBCT) images.
• Accurate estimation of interphase motion deformation fields is crucial for effective image-guided radiation therapy.
• Traditional motion compensation algorithms may struggle with significant artifacts, impacting diagnostic and therapeutic accuracy.

Purpose of the Study:

• To propose and evaluate a novel motion compensation reconstruction method using robust principal component analysis (RPCA) for CBCT.
• To reduce the influence of streak artifacts on the estimation of interphase motion deformation fields.
• To improve the overall quality of reconstructed CBCT images compared to existing methods.

Main Methods:

• A RPCA-based motion compensation reconstruction algorithm was developed, building upon the traditional MC-FDK algorithm.
• RPCA was employed to decompose CBCT images into low-rank and sparse components.
• Interphase motion deformation fields were estimated from low-rank images using the Horn and Schunck optical flow method to mitigate artifact impact.

Main Results:

• The proposed method demonstrated clearer tissue boundaries and reduced motion artifacts in reconstructed images compared to the traditional MC-FDK algorithm.
• Phantom data reconstruction showed significant improvements: 25.4% increase in Peak Signal-to-Noise Ratio (PSNR) and 7.6% in Structural Similarity Index Measure (SSIM) over MC-FDK.
• Further improvements were observed compared to the FDK algorithm, with PSNR increased by 37.9% and SSIM by 17.6%.

Conclusions:

• The proposed RPCA-based motion compensation reconstruction algorithm effectively reduces artifacts in CBCT images.
• The method achieves accurate estimation of inter-phase motion deformation fields.
• This technique enhances the quality of reconstructed CBCT images, offering potential benefits for clinical applications.