TY - GEN
T1 - Generating Dual-Energy Subtraction Soft-Tissue Images from Chest Radiographs via Bone Edge-Guided GAN
AU - Liu, Yunbi
AU - Liu, Mingxia
AU - Xi, Yuhua
AU - Qin, Genggeng
AU - Shen, Dinggang
AU - Yang, Wei
N1 - Funding Information:
Acknowledgements. Y. Liu, Y. Xi and W. Yang were partially supported by the National Natural Science Foundation of China (No. 81771916) and the Guangdong Provincial Key Laboratory of Medical Image Processing (No. 2014B-030301042). A part of this work was finished when Y. Liu was visiting the University of North Carolina at Chapel Hill.
Publisher Copyright:
© 2020, Springer Nature Switzerland AG.
PY - 2020
Y1 - 2020
N2 - Generating dual-energy subtraction (DES) soft-tissue images from chest radiographs (also called bone suppression) is an important task, as it improves the detection rates for lung nodules. Previous studies focus on generating DES-like soft-tissue images from CXRs through machine/deep learning techniques. However, they usually require tedious image processing steps for bone segmentation/delineation or ignore anatomical structure information (e.g., edges of ribs and clavicles) in CXRs. In this work, we propose a bone Edge-guided Generative Adversarial Network (EGAN) to generate DES-like soft-tissue images from conventional CXRs, which does not require human intervention and can explicitly use anatomical structure information of bones in CXRs. Specifically, the edges of ribs and clavicles in an input CXR were first detected by a trained fully convolutional network. Then, the edge probability map, as well as the original CXR image, are fed into a GAN model to generate the DES-like soft-tissue image, where the detected edge information is used as the prior knowledge to directly and specifically guide the image generation process. Experimental results on 504 subjects (each equipped with a CXR, a DES bone image, and a DES soft-tissue image) demonstrate that EGAN can produce DES-like soft-tissue images with high-quality and high-resolution, compared with classic deep learning methods. We further apply the trained EGAN to CXRs acquired by different types of X-ray machines in the public JSRT and NIH ChestXray 14 datasets, and our method can also produce visually appealing DES-like soft-tissue images.
AB - Generating dual-energy subtraction (DES) soft-tissue images from chest radiographs (also called bone suppression) is an important task, as it improves the detection rates for lung nodules. Previous studies focus on generating DES-like soft-tissue images from CXRs through machine/deep learning techniques. However, they usually require tedious image processing steps for bone segmentation/delineation or ignore anatomical structure information (e.g., edges of ribs and clavicles) in CXRs. In this work, we propose a bone Edge-guided Generative Adversarial Network (EGAN) to generate DES-like soft-tissue images from conventional CXRs, which does not require human intervention and can explicitly use anatomical structure information of bones in CXRs. Specifically, the edges of ribs and clavicles in an input CXR were first detected by a trained fully convolutional network. Then, the edge probability map, as well as the original CXR image, are fed into a GAN model to generate the DES-like soft-tissue image, where the detected edge information is used as the prior knowledge to directly and specifically guide the image generation process. Experimental results on 504 subjects (each equipped with a CXR, a DES bone image, and a DES soft-tissue image) demonstrate that EGAN can produce DES-like soft-tissue images with high-quality and high-resolution, compared with classic deep learning methods. We further apply the trained EGAN to CXRs acquired by different types of X-ray machines in the public JSRT and NIH ChestXray 14 datasets, and our method can also produce visually appealing DES-like soft-tissue images.
UR - http://www.scopus.com/inward/record.url?scp=85092714272&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-59713-9_65
DO - 10.1007/978-3-030-59713-9_65
M3 - Conference contribution
AN - SCOPUS:85092714272
SN - 9783030597122
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 678
EP - 687
BT - Medical Image Computing and Computer Assisted Intervention – MICCAI 2020 - 23rd International Conference, Proceedings
A2 - Martel, Anne L.
A2 - Abolmaesumi, Purang
A2 - Stoyanov, Danail
A2 - Mateus, Diana
A2 - Zuluaga, Maria A.
A2 - Zhou, S. Kevin
A2 - Racoceanu, Daniel
A2 - Joskowicz, Leo
PB - Springer Science and Business Media Deutschland GmbH
T2 - 23rd International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2020
Y2 - 4 October 2020 through 8 October 2020
ER -