Weighted simultaneous algebraic reconstruction technique for tomosynthesis imaging of objects with high-attenuation...

Area of Science:

• Medical Imaging
• Image Reconstruction
• Computational Imaging

Background:

• Tomosynthesis imaging is crucial for visualizing 3D structures but is prone to artifacts.
• High-attenuation features in objects can significantly degrade image quality due to limited angle artifacts.
• Existing reconstruction techniques often struggle to mitigate these artifacts effectively.

Purpose of the Study:

• To introduce a novel nonlinear weighting scheme into the backprojection step of SART.
• To reduce limited angle artifacts in tomosynthesis imaging, particularly for objects with high-attenuation features.
• To enhance the accuracy and clarity of reconstructed 3D images.

Main Methods:

• A nonlinear weighting scheme was integrated into the SART backprojection operation.
• An algorithm was developed to estimate and reduce the contribution of projections causing artifacts.
• A four-dimensional backprojected space representation and a dissimilarity measure were used to automatically calculate weighting coefficients.
• Individual weighting coefficients were computed for each angular view and voxel position.

Main Results:

• The proposed method was validated using real 3D tomosynthesis datasets (phantom, apple, bone, hand).
• Reconstructions showed a significant reduction in out-of-focus artifacts, lower standard deviation (STD), and narrower artifact spread function (ASF) compared to standard SART.
• The algorithm successfully reduced artifacts in point-like, long, and fine structures.

Conclusions:

• The developed algorithm is feasible for reducing tomosynthesis artifacts caused by high-attenuation features.
• Weighting coefficients are assigned automatically, eliminating the need for segmentation or tissue classification.
• The algorithm is adaptable to various iterative reconstruction methods and can be extended to computed tomography.