Abstract:Background Dual‐energy computed tomography (DECT) and material decomposition play vital roles in quantitative medical imaging. However, the decomposition process may suffer from significant noise amplification, leading to severely degraded image signal‐to‐noise ratios (SNRs). While existing iterative algorithms perform noise suppression using different image priors, these heuristic image priors cannot accurately represent the features of the target image manifold. Although deep learning‐based decomposition methods have been reported, these methods are in the supervised‐learning framework requiring paired data for training, which is not readily available in clinical settings. Purpose This work aims to develop an unsupervised‐learning framework with data‐measurement consistency for image‐domain material decomposition in DECT. Methods The proposed framework combines iterative decomposition and deep learning‐based image prior in a generative adversarial network (GAN) architecture. In the generator module, a data‐fidelity loss is introduced to enforce the measurement consistency in material decomposition. In the discriminator module, the discriminator is trained to differentiate the low‐noise material‐specific images from the high‐noise images. In this scheme, paired images of DECT and ground‐truth material‐specific images are not required for the model training. Once trained, the generator can perform image‐domain material decomposition with noise suppression in a single step. Results In the simulation studies of head and lung digital phantoms, the proposed method reduced the standard deviation (SD) in decomposed images by 97% and 91% from the values in direct inversion results. It also generated decomposed images with structural similarity index measures (SSIMs) greater than 0.95 against the ground truth. In the clinical head and lung patient studies, the proposed method suppressed the SD by 95% and 93% compared to the decomposed images of matrix inversion. Conclusions Since the invention of DECT, noise amplification during material decomposition has been one of the biggest challenges, impeding its quantitative use in clinical practice. The proposed method performs accurate material decomposition with efficient noise suppression. Furthermore, the proposed method is within an unsupervised‐learning framework, which does not require paired data for model training and resolves the issue of lack of ground‐truth data in clinical scenarios.

Prior image-based generative adversarial learning for multi-material decomposition in photon counting computed tomography

Material decomposition from photon-counting CT using a convolutional neural network and energy-integrating CT training labels

Deep Learning for Material Decomposition in Photon-Counting CT

Image‐domain material decomposition for dual‐energy CT using unsupervised learning with data‐fidelity loss

PMS-GAN: Parallel Multi-Stream Generative Adversarial Network for Multi-Material Decomposition in Spectral Computed Tomography.

A material decomposition method for dual-energy CT via dual interactive Wasserstein generative adversarial networks

An Image-Based Prior Knowledge-Free Approach for a Multi-Material Decomposition in Photon-Counting Computed Tomography

Image-Domain Material Decomposition for Dual-energy CT using Unsupervised Learning with Data-fidelity Loss

Accurate Multi-Material Decomposition in Dual-Energy CT: A Phantom Study

MMD‐Net: Image domain multi‐material decomposition network for dual‐energy CT imaging

Quantitative dual-energy CBCT imaging with deep triple-material decomposition

Robust Multimaterial Decomposition of Spectral CT Using Convolutional Neural Networks

Material decomposition of spectral CT images via attention-based global convolutional generative adversarial network

Image Decomposition Algorithm For Dual-Energy Computed Tomography Via Fully Convolutional Network

A Multi-Energy Material Decomposition Method for Spectral CT Using Neural Network

Feasibility study of three-material decomposition in dual-energy cone-beam CT imaging with deep learning

Material decomposition using dual-energy CT with unsupervised learning

Image Domain Multi-Material Decomposition Using Single Energy CT.

DECT-MULTRA: Dual-Energy CT Image Decomposition With Learned Mixed Material Models and Efficient Clustering

Direct Dual Energy CT Material Decomposition Using Noise2Noise Prior

Dual-Energy Ct Reconstruction Using Deep Mutual-Domain Knowledge For Basis Decomposition And Denoising