Simultaneous 4D-CBCT reconstruction with sliding motion constraint.

Area of Science:

• Medical Imaging
• Image Reconstruction
• Computational Anatomy

Background:

• Current 4D-cone beam CT (CBCT) reconstruction methods often use isotropic deformable vector fields (DVFs), which are inadequate for complex motions like organ sliding.
• Inaccurate motion modeling in 4D-CBCT can lead to artifacts and reduced diagnostic accuracy, particularly in scenarios with inter-organ sliding.

Purpose of the Study:

• To develop and validate an anisotropic motion modeling scheme for 4D-CBCT reconstruction that accounts for organ boundary sliding motion.
• To improve the accuracy of motion compensation and the quality of reconstructed 4D-CBCT images by integrating local motion information.

Main Methods:

• A modified simultaneous algebraic reconstruction technique (mSART) was employed for initial CBCT reconstruction.
• Sliding motion was estimated by matching projections and incorporated into DVF optimization using boundary normal vectors.
• Anisotropic regularization adaptively smoothed the DVF, allowing sliding motion along boundary tangent directions.

Main Results:

• The proposed method significantly reduced reconstruction errors (e.g., 7.4% RE for NCAT phantom vs. 8.9% without sliding motion).
• Motion trajectory errors were substantially decreased (e.g., 0.5 mm RMSE vs. 3.5 mm for NCAT phantom).
• Improved lung boundary similarity was observed in both phantom and patient data, indicated by higher Dice coefficients.

Conclusions:

• A novel sliding motion-compensated 4D-CBCT reconstruction method with anisotropic motion modeling was successfully developed.
• The developed scheme demonstrated enhanced accuracy in both motion modeling and 4D-CBCT reconstruction quality in phantom and patient studies.