A factorization approach for cone-beam reconstruction on a circular short-scan

Area of Science:

• Medical Imaging
• Computational Imaging
• Image Reconstruction

Background:

• Cone-beam (CB) imaging is crucial for 3-D reconstruction.
• Partial circular scans present reconstruction challenges.
• Existing algorithms for circular short-scan CB reconstruction have limitations.

Purpose of the Study:

• Introduce a novel algorithm for 3-D image reconstruction from partial circular cone-beam scans.
• Address the computational complexity and accuracy of 3-D reconstruction in limited-scan scenarios.

Main Methods:

• Developed an exact factorization of the 3-D reconstruction problem into independent 2-D inversion problems.
• Solved each 2-D problem using a projected steepest descent iteration scheme.
• Evaluated the algorithm using simulated cone-beam data (FORBILD head and disk phantoms) with and without noise.

Main Results:

• The proposed factorization algorithm successfully reconstructs 3-D images from partial circular scans.
• Numerical evaluations demonstrate the algorithm's performance and the impact of reconstruction parameters.
• Reconstruction results show comparable or improved image quality compared to state-of-the-art methods for circular short-scan.

Conclusions:

• The novel factorization approach provides an effective solution for 3-D cone-beam image reconstruction from partial scans.
• The algorithm offers a promising alternative for applications requiring efficient and high-quality 3-D imaging with reduced scan times.