Total variation-based neutron computed tomography

Area of Science:

• Physics
• Computer Science
• Materials Science

Background:

• Neutron computed tomography (NCT) is crucial for non-destructive material analysis.
• Traditional reconstruction methods like filtered back projection suffer from artifacts with limited angular data.
• Inverse problem formulations offer a path to improved reconstruction quality.

Purpose of the Study:

• To develop an efficient algorithm for NCT reconstruction from under-sampled angular measurements.
• To suppress high-frequency artifacts common in NCT.
• To enable accurate reconstruction with significantly fewer projection angles.

Main Methods:

• Formulation of the NCT reconstruction as an inverse problem.
• Application of a total variation (TV) penalty to suppress artifacts.
• Implementation of a split Bregman algorithm for efficient computation.
• Utilizing very inexact approximate linear solvers to reduce computational cost.

Main Results:

• The split Bregman algorithm effectively suppresses artifacts in low-angular sampling NCT.
• The method demonstrates the ability to capture small features even with extremely limited angles.
• Successful application shown on both synthetic and real neutron source data.
• Significant reduction in computational cost achieved through inexact solvers.

Conclusions:

• The proposed algorithm provides an effective solution for NCT reconstruction under severe angular undersampling.
• This approach enhances image quality and feature visibility in challenging data acquisition scenarios.
• The method holds promise for improving the efficiency and applicability of NCT.