Superiorized algorithm for reconstruction of CT images from sparse-view and limited-angle polyenergetic data.

Area of Science:

• Medical Imaging
• Computational Imaging
• Image Reconstruction

Background:

• Iterative algorithms in CT image reconstruction often use total variation (TV) penalties for undersampled data.
• Superiorization is a heuristic that enhances iterative algorithms to improve objective functions.
• Existing superiorization methods typically assume a monoenergetic X-ray spectrum, leading to beam hardening artifacts in polyenergetic CT.

Purpose of the Study:

• To adapt the superiorization heuristic for iterative CT image reconstruction using polyenergetic X-ray data.
• To investigate the effectiveness of superiorization with total variation (TV) and anisotropic TV (ATV) penalties in reducing artifacts.
• To evaluate the performance in sparse-view and limited-angle CT scenarios.

Main Methods:

• Superiorization of an iterative algorithm for polyenergetic CT data reconstruction.
• Application of both TV and anisotropic TV (ATV) penalties.
• Numerical phantom experiments simulating sparse-view and limited-angle data acquisition.

Main Results:

• The superiorized algorithm produced solutions compatible with problem constraints.
• Significantly reduced TV and ATV values were achieved compared to the original algorithm.
• The method effectively reduced sparse-view and limited-angle artifacts.
• Images were largely free of beam hardening artifacts.

Conclusions:

• Superiorization is effective for iterative CT reconstruction with polyenergetic data.
• The approach successfully mitigates artifacts from undersampling and beam hardening.
• This method offers improved image quality for challenging CT acquisition scenarios.