Non-convexly constrained image reconstruction from nonlinear tomographic X-ray measurements

Area of Science:

• Medical imaging
• Computational imaging
• Materials science

Background:

• Polychromatic X-ray sources in tomography introduce nonlinear transmission effects.
• Standard tomographic reconstruction methods often ignore these nonlinearities, leading to artifacts.
• Artifacts are severe in multi-material objects with diverse X-ray attenuation.

Purpose of the Study:

• To investigate nonlinear X-ray transmission models for tomographic reconstruction.
• To develop advanced algorithms incorporating non-convex constraints.
• To enable accurate volumetric data reconstruction with limited measurements.

Main Methods:

• Development of a nonlinear conjugate gradient iterative hard thresholding algorithm.
• Application of non-convex constraints to impose prior modeling assumptions.
• Testing reconstruction performance with limited and multi-material datasets.

Main Results:

• Demonstrated improved volumetric data reconstruction using nonlinear models.
• Successfully mitigated artifacts caused by X-ray nonlinearities.
• Showcased the effectiveness of non-convex constraints in limited data scenarios.

Conclusions:

• Nonlinear X-ray transmission models are crucial for accurate tomography.
• The proposed algorithm effectively handles multi-material imaging challenges.
• Non-convex constraints enhance reconstruction robustness and data efficiency.