Optimization-based image reconstruction regularized with inter-spectral structural similarity for limited-angle...

Area of Science:

• Medical Imaging
• Radiology
• Computational Imaging

Background:

• Limited-angle dual-energy (DE) cone-beam CT (CBCT) offers fast, low-dose imaging on existing scanners.
• Clinical use is restricted by difficult image reconstruction from limited projection angles.
• Existing methods require spectral measurements or large paired datasets, limiting practicality.

Purpose of the Study:

• To develop a practical solution for image reconstruction in limited-angle DE-CBCT.
• To facilitate clinical applications of fast and low-dose DE-CBCT.
• To reduce artifacts in reconstructed images without specialized hardware.

Main Methods:

• Integrated an inter-spectral structural similarity-based regularization into iterative image reconstruction.
• Enforced similarity between DE images to reduce limited-angle artifacts.
• Evaluated the method using physical and digital phantoms.

Main Results:

• Achieved accurate image reconstruction without residual artifacts from limited-angle DE-CBCT data.
• Reduced mean absolute error from 309/290 HU to 14/20 HU in digital phantoms.
• Increased peak signal-to-noise ratio and improved structural similarity index measurements.

Conclusions:

• The method efficiently reduces limited-angle artifacts in DE-CBCT reconstruction.
• Enables quantitative DE-CBCT with acquisition times and radiation doses comparable to single-energy scans.
• Significant potential to advance clinical applications in image-guided interventions and radiation therapy.