A Locally Weighted Linear Regression Look-Up Table-Based Iterative Reconstruction Method for Dual Spectral CT

Area of Science:

• Medical Imaging
• Computational Imaging
• Materials Science

Background:

• Dual-spectral CT (DSCT) offers superior material distinguishability compared to traditional CT.
• Accurate forward-projection function modeling is critical for iterative DSCT algorithms but analytically challenging.
• DSCT has significant potential in both industrial and medical applications.

Purpose of the Study:

• To develop an iterative reconstruction method for DSCT that accurately models forward-projection functions.
• To address the difficulty of analytically providing accurate forward-projection functions in DSCT.
• To improve the quality of reconstructed DSCT images.

Main Methods:

• A locally weighted linear regression look-up table (LWLR-LUT) based iterative reconstruction method was proposed.
• LWLR was used to establish look-up tables (LUTs) for forward-projection functions via calibration phantoms.
• The method implicitly accounts for scattered radiation and fits forward-projection functions locally.

Main Results:

• The LWLR-LUT method achieved highly accurate polychromatic forward-projection functions.
• Image quality was significantly improved for both scattering-free and scattering projections.
• Numerical simulations and real data experiments validated the method's effectiveness.

Conclusions:

• The proposed LWLR-LUT method is simple, practical, and effective for DSCT.
• It achieves good material decomposition for objects with complex structures.
• Calibration through simple phantoms is sufficient for the method's success.