Image-domain multi-material decomposition for dual-energy CT based on correlation and sparsity of material images

TL;DR

This paper introduces a novel image-domain multi-material decomposition method for dual-energy CT that leverages correlations and sparsity among material images, improving accuracy and efficiency in reconstructing multiple basis materials.

Contribution

The proposed method incorporates prior information about common edges and sparsity in material images, addressing limitations of previous approaches that neglect inter-material relations and are computationally intensive.

Findings

Enhanced accuracy in multi-material reconstruction.

Reduced computational complexity compared to existing methods.

Effective utilization of material image correlations and sparsity.

Abstract

Dual energy CT (DECT) enhances tissue characterization because it can produce images of basis materials such as soft-tissue and bone. DECT is of great interest in applications to medical imaging, security inspection and nondestructive testing. Theoretically, two materials with different linear attenuation coefficients can be accurately reconstructed using DECT technique. However, the ability to reconstruct three or more basis materials is clinically and industrially important. Under the assumption that there are at most three materials in each pixel, there are a few methods that estimate multiple material images from DECT measurements by enforcing sum-to-one and a box constraint ([0 1]) derived from both the volume and mass conservation assumption. The recently proposed image-domain multi-material decomposition (MMD) method introduces edge-preserving regularization for each material image which neglects the relations among material images, and enforced the assumption that there are at most three materials in each pixel using a time-consuming loop over all possible material-triplet in each iteration of optimizing its cost function. We propose a new image-domain MMD method for DECT that considers the prior information that different material images have common edges and encourages sparsity of material composition in each pixel using regularization.