Technical note: algebraic iterative image reconstruction using a cylindrical image grid for tetrahedron beam computed...

Area of Science:

• Medical Imaging
• Computational Imaging
• Image Reconstruction Algorithms

Background:

• Cone beam computed tomography (CBCT) and tetrahedron beam computed tomography (TBCT) face challenges with scatter and detector issues.
• Iterative algebraic reconstruction algorithms (IARAs) can reduce artifacts in CBCT but are computationally intensive, limiting clinical use.
• Traditional Cartesian grids are computationally expensive for volumetric CT methods with circular scanning trajectories.

Purpose of the Study:

• To accelerate iterative algebraic reconstruction algorithms by employing a cylindrical image grid.
• To leverage the inherent symmetries of cylindrical geometry for improved computational efficiency in CT reconstruction.
• To evaluate the accuracy and performance of a cylindrical voxelization method combined with the simultaneous algebraic reconstruction technique (SART).

Main Methods:

• Developed a cylindrical voxelization method on a cylindrical grid, suitable for circular scanning trajectories in CT.
• Integrated the cylindrical voxelization with the simultaneous algebraic reconstruction technique (SART).
• Validated the method using 2D/3D numerical phantoms and a patient CT image, comparing results against Cartesian reconstructions using figures of merit (FOM).

Main Results:

• Successfully implemented cylindrical voxelization with SART, demonstrating comparable accuracy to Cartesian reconstructions.
• Achieved superior accuracy in 3D reconstructions using the cylindrical method compared to Cartesian methods.
• Reported a significant speedup factor of approximately 40 and a reduction in system matrix storage by two orders of magnitude.

Conclusions:

• Cylindrical image grid reconstruction for TBCT significantly enhances reconstruction speed and reduces storage requirements.
• The cylindrical method maintains image quality comparable to Cartesian voxelization.
• This approach offers a more compact and efficient solution for CT image reconstruction, facilitating clinical implementation.