Fast iterative reconstruction of data in full interior tomography

Area of Science:

• Medical Imaging
• Computational Science
• Materials Science

Background:

• Full interior tomography (FINT) presents reconstruction challenges due to limited field-of-view and unknown object support or attenuation.
• Existing analytical techniques struggle with underconstrained FINT datasets, impacting downstream analysis.

Purpose of the Study:

• Introduce novel iterative reconstruction strategies for FINT data.
• Address limitations of current methods for FINT reconstruction.
• Enhance the quality of reconstructed FINT images for improved analysis.

Main Methods:

• Developed two iterative reconstruction strategies: data edge-padding and intermediate virtual sinogram creation.
• Validated methods using the alternate direction method of multipliers plug-and-play with gridding projectors.
• Achieved significant speed-up (three orders of magnitude) compared to standard real-space operators.

Main Results:

• Both proposed methods substantially improve reconstruction quality for underconstrained FINT datasets.
• Demonstrated superior performance compared to analytical techniques.
• Validated on synchrotron-based X-ray tomographic microscopy datasets of mouse lung alveoli.

Conclusions:

• The novel iterative strategies effectively address FINT reconstruction challenges.
• Improved reconstruction quality facilitates subsequent post-processing steps.
• These methods offer a significant advancement for analyzing complex tomographic data.