Deep Learning-Based Computed Tomography Image Standardization to Improve Generalizability of Deep Learning-Based Hepatic Segmentation

被引:6
作者
Lee, Seul Bi [1 ,2 ]
Hong, Youngtaek [3 ]
Cho, Yeon Jin [1 ,2 ,7 ]
Jeong, Dawun [3 ,4 ]
Lee, Jina [3 ,4 ]
Yoon, Soon Ho [1 ,2 ,5 ]
Lee, Seunghyun [1 ,2 ]
Choi, Young Hun [1 ,2 ]
Cheon, Jung-Eun [1 ,2 ,6 ]
机构
[1] Seoul Natl Univ Hosp, Dept Radiol, Seoul, South Korea
[2] Seoul Natl Univ, Dept Radiol, Coll Med, Seoul, South Korea
[3] Yonsei Univ, AI R&D Ctr, CONNECT, Coll Med, Seoul, South Korea
[4] Yonsei Univ, Brain Korea 21 PLUS Project Med Sci, Seoul, South Korea
[5] MEDICALIP Co Ltd, Seoul, South Korea
[6] Seoul Natl Univ, Inst Radiat Med, Med Res Ctr, Seoul, South Korea
[7] Seoul Natl Univ Hosp, Dept Radiol, 101 Daehak Ro, Seoul 03080, South Korea
基金
新加坡国家研究基金会;
关键词
Artificial intelligence; Automated segmentation; Image conversion; Quality control; Reproducibility; DATA AUGMENTATION; GUIDE;
D O I
10.3348/kjr.2022.0588
中图分类号
R8 [特种医学]; R445 [影像诊断学];
学科分类号
1002 ; 100207 ; 1009 ;
摘要
Objective: We aimed to investigate whether image standardization using deep learning-based computed tomography (CT) image conversion would improve the performance of deep learning-based automated hepatic segmentation across various reconstruction methods. Materials and Methods: We collected contrast-enhanced dual-energy CT of the abdomen that was obtained using various reconstruction methods, including filtered back projection, iterative reconstruction, optimum contrast, and monoenergetic images with 40, 60, and 80 keV. A deep learning based image conversion algorithm was developed to standardize the CT images using 142 CT examinations (128 for training and 14 for tuning). A separate set of 43 CT examinations from 42 patients (mean age, 10.1 years) was used as the test data. A commercial software program (MEDIP PRO v2.0.0.0, MEDICALIP Co. Ltd.) based on 2D U-NET was used to create liver segmentation masks with liver volume. The original 80 keV images were used as the ground truth. We used the paired t-test to compare the segmentation performance in the Dice similarity coefficient (DSC) and difference ratio of the liver volume relative to the ground truth volume before and after image standardization. The concordance correlation coefficient (CCC) was used to assess the agreement between the segmented liver volume and ground-truth volume. Results: The original CT images showed variable and poor segmentation performances. The standardized images achieved significantly higher DSCs for liver segmentation than the original images (DSC [original, 5.40%-91.27%] vs. [standardized, 93.16%-96.74%], all P < 0.001). The difference ratio of liver volume also decreased significantly after image conversion (original, 9.84%-91.37% vs. standardized, 1.99%-4.41%). In all protocols, CCCs improved after image conversion (original,-0.006-0.964 vs. standardized, 0.990-0.998). Conclusion: Deep learning-based CT image standardization can improve the performance of automated hepatic segmentation using CT images reconstructed using various methods. Deep learning-based CT image conversion may have the potential to improve the generalizability of the segmentation network.
引用
收藏
页码:294 / 304
页数:11
相关论文
共 32 条
[11]   Perceptual Losses for Real-Time Style Transfer and Super-Resolution [J].
Johnson, Justin ;
Alahi, Alexandre ;
Li Fei-Fei .
COMPUTER VISION - ECCV 2016, PT II, 2016, 9906 :694-711
[12]   A deep learning algorithm may automate intracranial aneurysm detection on MR angiography with high diagnostic performance [J].
Joo, Bio ;
Ahn, Sung Soo ;
Yoon, Pyeong Ho ;
Bae, Sohi ;
Sohn, Beomseok ;
Lee, Yong Eun ;
Bae, Jun Ho ;
Park, Moo Sung ;
Choi, Hyun Seok ;
Lee, Seung-Koo .
EUROPEAN RADIOLOGY, 2020, 30 (11) :5785-5793
[13]  
Kawaguchi K, 2020, Arxiv, DOI [arXiv:1710.05468, DOI 10.48550/ARXIV.1710.05468]
[14]  
King DB, 2015, ACS SYM SER, V1214, P1, DOI 10.1021/bk-2015-1214.ch001
[15]   Expert-level detection of acute intracranial hemorrhage on head computed tomography using deep learning [J].
Kuo, Weicheng ;
Hane, Christian ;
Mukherjee, Pratik ;
Malik, Jitendra ;
Yuh, Esther L. .
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 2019, 116 (45) :22737-22745
[16]  
Ledig C., Photo realistic single
[17]   Deep Learning-Based Image Conversion Improves the Reproducibility of Computed Tomography Radiomics Features A Phantom Study [J].
Lee, Seul Bi ;
Cho, Yeon Jin ;
Hong, Youngtaek ;
Jeong, Dawun ;
Lee, Jina ;
Kim, Soo-Hyun ;
Lee, Seunghyun ;
Choi, Young Hun .
INVESTIGATIVE RADIOLOGY, 2022, 57 (05) :308-317
[18]   A CONCORDANCE CORRELATION-COEFFICIENT TO EVALUATE REPRODUCIBILITY [J].
LIN, LI .
BIOMETRICS, 1989, 45 (01) :255-268
[19]   Deep Learning in Radiology [J].
McBee, Morgan P. ;
Awan, Omer A. ;
Colucci, Andrew T. ;
Ghobadi, Comeron W. ;
Kadom, Nadja ;
Kansagra, Akash P. ;
Tridandapani, Srini ;
Auffermann, William F. .
ACADEMIC RADIOLOGY, 2018, 25 (11) :1472-1480
[20]  
Onofrey JA, 2019, I S BIOMED IMAGING, P348, DOI [10.1109/ISBI.2019.8759295, 10.1109/isbi.2019.8759295]