TICMR: Total Image Constrained Material Reconstruction via Nonlocal Total Variation Regularization for Spectral CT

Area of Science:

• Medical Imaging
• Computational Imaging
• Materials Science

Background:

• Spectral CT provides material-specific information but often suffers from low signal-to-noise ratio (SNR).
• Reconstructing material composition directly from spectral data is challenging due to inherent noise.
• Existing methods often involve separate image reconstruction and material decomposition steps, which can be suboptimal.

Purpose of the Study:

• To develop a novel material reconstruction method for spectral CT that leverages both spectral information and high-SNR total images.
• To improve the signal-to-noise ratio (SNR) and image quality of material decomposition in spectral CT.
• To enable direct reconstruction of material compositions, synergizing material decomposition and image reconstruction.

Main Methods:

• Proposed Total Image Constrained Material Reconstruction (TICMR) method.
• Utilized spectrally-integrated measurements to reconstruct a high-SNR total image.
• Incorporated nonlocal total variation (NLTV) regularization, guided by the high-SNR total image, into an iterative material reconstruction scheme.
• Employed the alternating direction method of multipliers (ADMM) for solving the optimization problem.

Main Results:

• TICMR successfully reconstructed material compositions with improved SNR.
• The method demonstrated enhanced image quality, including better contrast-to-noise ratio (CNR) and spatial resolution.
• Validated through both simulated and experimental spectral CT data.

Conclusions:

• TICMR offers a significant advancement in spectral CT material reconstruction.
• The proposed method effectively balances spectral information utilization with noise reduction.
• TICMR provides superior performance compared to traditional filtered back-projection (FBP) and total-variation-based iterative methods.