4D cone-beam CT reconstruction using multi-organ meshes for sliding motion modeling

Area of Science:

• Medical Imaging
• Computational Anatomy
• Radiotherapy Physics

Background:

• 4D cone-beam CT (4D-CBCT) is crucial for radiotherapy, requiring accurate motion estimation.
• Existing simultaneous motion estimation and image reconstruction (SMEIR) methods use voxel grids, limiting efficiency and accuracy.
• Global smoothness regularization in original SMEIR struggles with inter-organ motion discontinuities.

Purpose of the Study:

• To enhance the computational efficiency and motion estimation accuracy of SMEIR for 4D-CBCT.
• To develop a multi-organ meshing model for improved 4D-CBCT reconstruction.
• To accurately capture organ motion, including discontinuities, for better tumor visualization and tracking.

Main Methods:

• Developed a multi-organ meshing model with feature-based adaptive meshes to reduce unknowns in deformation vector field (DVF) estimation.
• Incorporated explicit consideration of motion discontinuities between organs during respiration.
• Implemented a GPU-based parallel computing strategy for enhanced computational efficiency.
• Evaluated the method on synthetic phantoms, a 4D NCAT phantom, and four lung cancer patients.

Main Results:

• The multi-organ mesh-based SMEIR significantly improved motion estimation accuracy and computational efficiency compared to conventional methods.
• Accurate visualization and motion estimation of tumors, especially on organ boundaries, were achieved.
• The proposed method outperformed Feldkamp-Davis-Kress, iterative total variation minimization, original SMEIR, and single meshing techniques.

Conclusions:

• The multi-organ meshing approach represents a significant advancement for SMEIR in 4D-CBCT.
• This method offers improved accuracy and efficiency for lung cancer radiotherapy guidance.
• The strategy effectively handles complex organ motion and discontinuities, crucial for precise treatment delivery.