NCCT-CECT image synthesizers and their application to pulmonary vessel segmentation

Pang, Haowen1,2; Qi, Shouliang1,2; Wu, Yanan1,2; Wang, Meihuan1,2; Li, Chen1,2; Sun, Yu1,3; Qian, Wei1,2; Tang, Guoyan4; Xu, Jiaxuan4; Liang, Zhenyu4; Chen, Rongchang5

2023-04-01

10.1016/j.cmpb.2023.107389

COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE   Impact Factor & Quartile

0169-2607

1872-7565

Background and objectives: Non-contrast CT (NCCT) and contrast-enhanced CT (CECT) are important diag-nostic tools with distinct features and applications for chest diseases. We developed two synthesizers for the mutual synthesis of NCCT and CECT and evaluated their applications.Methods: Two synthesizers (S1 and S2) were proposed based on a generative adversarial network. S1 generated synthetic CECT (SynCECT) from NCCT and S2 generated synthetic NCCT (SynNCCT) from CECT. A new training procedure for synthesizers was proposed. Initially, the synthesizers were pretrained using self-supervised learning (SSL) and dual-energy CT (DECT) and then fine-tuned using the registered NCCT and CECT images. Pulmonary vessel segmentation from NCCT was used as an example to demonstrate the effectiveness of the synthesizers. Two strategies (ST1 and ST2) were proposed for pulmonary vessel segmentation. In ST1, CECT images were used to train a segmentation model (Model-CECT), NCCT im-ages were converted to SynCECT through S1, and SynCECT was input to Model-CECT for testing. In ST2, CECT data were converted to SynNCCT through S2. SynNCCT and CECT-based annotations were used to train an additional model (Model-NCCT), and NCCT was input to Model-NCCT for testing. Three datasets, D1 (40 paired CTs), D2 (14 NCCTs and 14 CECTs), and D3 (49 paired DECTs), were used to evaluate the synthesizers and strategies.Results: For S1, the mean absolute error (MAE), mean squared error (MSE), peak signal-to-noise ratio (PSNR), and structural similarity index (SSIM) were 14.60 +/- 2.19, 1644 +/- 890, 34.34 +/- 1.91, and 0.94 +/- 0.02, respectively. For S2, they were 12.52 +/- 2.59, 1460 +/- 922, 35.08 +/- 2.35, and 0.95 +/- 0.02, respectively. Our synthesizers outperformed the counterparts of CycleGAN, Pix2Pix, and Pix2PixHD. The results of ablation studies on SSL pretraining, DECT pretraining, and fine-tuning showed that performance worsened (for example, for S1, MAE increased to 16.53 +/- 3.10, 17.98 +/- 3.10, and 20.57 +/- 3.75, respectively). Model-NCCT and Model-CECT achieved dice similarity coefficients (DSC) of 0.77 and 0.86 on D1 and 0.77 and 0.72 on D2, respectively.Conclusions: The proposed synthesizers realized mutual and high-quality synthesis between NCCT and CECT images; the training procedures, including SSL pretraining, DECT pretraining, and fine-tuning, were critical to their effectiveness. The results demonstrated the usefulness of synthesizers for pulmonary ves -sel segmentation from NCCT images.(c) 2023 Elsevier B.V. All rights reserved.

Medical image synthesis Generative adversarial network Computed tomography Pulmonary vessel segmentation

SCI

English

Corresponding

National Natural Science Foundation of China["820720 08","82270 044","62271131"] ; Fundamental Research Funds for the Central Universities["N2119010","N2224001-10"] ; Natural Science Founda- tion of Liaoning Province["2021-YGJC-21","2020-BS-049"]

Computer Science ; Engineering ; Medical Informatics

Computer Science, Interdisciplinary Applications ; Computer Science, Theory & Methods ; Engineering, Biomedical ; Medical Informatics

WOS:000931686600001

ELSEVIER IRELAND LTD

COMPUTER SCIENCE

Web of Science

1.Northeastern Univ, Coll Med & Biol Informat Engn, Shenyang, Peoples R China
2.Northeastern Univ, Key Lab Intelligent Comp Med Image, Minist Educ, Shenyang, Peoples R China
3.Gen Hosp Northern Theater Command, Dept Radiol, Shenyang, Peoples R China
4.Guangzhou Med Univ, Affiliated Hosp 1, Natl Clin Res Ctr Resp Dis, Natl Ctr Resp Med,State Key Lab Resp Dis,Guangzhou, Guangzhou, Peoples R China
5.Jinan Univ, South Univ Sci & Technol China, Shenzhen Peoples Hosp, Shenzhen Inst Resp Dis,Key Lab Resp Dis Shenzhen,A, Shenzhen, Peoples R China

Pang, Haowen,Qi, Shouliang,Wu, Yanan,et al. NCCT-CECT image synthesizers and their application to pulmonary vessel segmentation[J]. COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE,2023,231.

Pang, Haowen.,Qi, Shouliang.,Wu, Yanan.,Wang, Meihuan.,Li, Chen.,...&Chen, Rongchang.(2023).NCCT-CECT image synthesizers and their application to pulmonary vessel segmentation.COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE,231.

Pang, Haowen,et al."NCCT-CECT image synthesizers and their application to pulmonary vessel segmentation".COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE 231(2023).