A Spectral CT Method to Directly Estimate Basis Material Maps From Experimental Photon-Counting Data.

Area of Science:

• Medical Imaging
• Computational Imaging
• Materials Science

Background:

• Spectral CT offers improved material differentiation over conventional CT.
• Accurate material decomposition requires precise spectral information and robust reconstruction algorithms.
• Existing methods often struggle with nonlinear detector effects and spectral uncertainties.

Purpose of the Study:

• To develop and validate a spectral CT reconstruction method for accurate basis material map estimation.
• To model nonlinear X-ray detection and incorporate convex constraints for improved image quality.
• To empirically estimate effective energy-window spectra and optimize spectral response for artifact reduction.

Main Methods:

• Constrained one-step spectral CT reconstruction (cOSSCIR) optimization.
• Empirical estimation of effective energy-window spectra via calibration.
• Joint optimization of spectral amplitudes and basis material maps.
• Validation approach for constraint parameter selection.

Main Results:

• Successful reconstruction of basis material maps in simulations and experiments.
• Achieved <1% error in simulations and low errors (0.5-4%) for specific materials in experiments.
• Demonstrated significant reduction in errors for Teflon region compared to methods without spectral estimation (-24%/126% vs. 8%/31%).
• Reduced ring artifacts and improved material decomposition angle accuracy (within 1.3 degrees).

Conclusions:

• The proposed spectral CT method with empirical spectral modeling and cOSSCIR is effective for accurate material decomposition.
• Joint spectral and material map estimation reduces artifacts and enhances quantitative accuracy.
• The developed validation procedure enables automated parameter selection for clinical feasibility.