ADN: Artifact Disentanglement Network for Unsupervised Metal Artifact Reduction

• Published in: IEEE transactions on medical imaging Vol. 39; no. 3; pp. 634 - 643
• Main Authors: Liao, Haofu, Lin, Wei-An, Zhou, S. Kevin, Luo, Jiebo
• Format: Journal Article
• Language: English
• Published: United States IEEE 01.03.2020; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Artificial neural networks; Computed tomography; Computer simulation; Datasets; Decoding; Humans; Image enhancement/restoration (noise and artifact reduction); Image Processing, Computer-Assisted - methods; Image reconstruction; Machine Learning; Mars; Medical imaging; Metals; Metals - isolation & purification; neural network; Neural networks; Neural Networks, Computer; Reduction (metal working); Source code; Synthesis; Therapeutic applications; Tomography, X-Ray Computed - methods; Training; X-ray imaging
• ISSN: 0278-0062; 1558-254X; 1558-254X
• DOI: 10.1109/TMI.2019.2933425

Current deep neural network based approaches to computed tomography (CT) metal artifact reduction (MAR) are supervised methods that rely on synthesized metal artifacts for training. However, as synthesized data may not accurately simulate the underlying physical mechanisms of CT imaging, the supervised methods often generalize poorly to clinical applications. To address this problem, we propose, to the best of our knowledge, the first unsupervised learning approach to MAR. Specifically, we introduce a novel artifact disentanglement network that disentangles the metal artifacts from CT images in the latent space. It supports different forms of generations (artifact reduction, artifact transfer, and self-reconstruction, etc.) with specialized loss functions to obviate the need for supervision with synthesized data. Extensive experiments show that when applied to a synthesized dataset, our method addresses metal artifacts significantly better than the existing unsupervised models designed for natural image-to-image translation problems, and achieves comparable performance to existing supervised models for MAR. When applied to clinical datasets, our method demonstrates better generalization ability over the supervised models. The source code of this paper is publicly available at https:// github.com/liaohaofu/adn.