Robust Iterative Methods: Convergence and Applications to Proton Computed Tomography.

Area of Science:

• Medical Imaging
• Computational Physics

Background:

• Robust methods like Tikhonov regularization are established for small, dense matrix problems.
• Their application to large-scale iterative methods in particle imaging, particularly pCT, is recent.

Purpose of the Study:

• To demonstrate the promise of robust methods in proton/ion computed tomography (pCT).
• To prove the convergence of robust methods for pCT image reconstruction.
• To highlight benefits for reconstruction in uncertain systems.

Main Methods:

• Application of robust methods (e.g., Tikhonov regularization, Bounded data uncertainty) to large-scale iterative algorithms.
• Analysis of image reconstruction in the presence of measurement and path errors.
• Evaluation of performance in low-dose and incomplete data scenarios.

Main Results:

• Robust methods show significant promise for pCT image reconstruction.
• Convergence of these robust methods has been mathematically proven for the first time.
• Improved accuracy in reconstructions with proton/ion energy measurement errors and path uncertainties.

Conclusions:

• Robust methods offer substantial advantages for pCT, particularly in handling data uncertainties.
• The proven convergence of robust methods is a key benefit for reliable pCT reconstruction.
• These methods enable high-quality imaging even with limited data or inherent system errors.