Region-of-interest tomography using filtered backprojection: assessing the practical limits

Area of Science:

• X-ray computed tomography
• Image reconstruction
• Materials science

Background:

• Reconstructing high-resolution images of small regions from larger samples is a common challenge in various scientific fields.
• Existing methods may require complex algorithms or specialized equipment, limiting practical application.

Purpose of the Study:

• To investigate the practical limits of high-resolution region of interest (ROI) tomography from large samples.
• To evaluate the impact of sinogram truncation on image reconstruction fidelity.
• To assess simple strategies for mitigating artifacts introduced by truncation.

Main Methods:

• Simulated ROI tomography by artificially truncating sinograms of whole sample scans.
• Applied a filtered back projection algorithm with lateral sinogram extension.
• Evaluated the influence of truncation degree, projection count, and ROI position across diverse sample types.

Main Results:

• Feature detection remains largely unaffected by truncation for a wide range of objects.
• Excellent image quality is achievable when projection numbers are based on the required pixels for the whole sample at the desired resolution.
• Simple strategies effectively minimize artifacts without necessitating complex reconstruction algorithms or specialized hardware.

Conclusions:

• High-resolution ROI tomography is feasible with significant sinogram truncation.
• Sophisticated reconstruction strategies are often unnecessary for achieving high-quality images.
• Optimizing projection count based on desired resolution is key to successful reconstruction.