Deep learning with ultrasonography: automated classification of liver fibrosis using a deep convolutional neural network

Jeong Hyun Lee, Ijin Joo, Tae Wook Kang, Yong Han Paik, Dong Hyun Sinn, Sang Yun Ha, Kyunga Kim, Choonghwan Choi, Gunwoo Lee, Jonghyon Yi, Won Chul Bang

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Objectives: The aim of this study was to develop a deep convolutional neural network (DCNN) for the prediction of the METAVIR score using B-mode ultrasonography images. Methods: Datasets from two tertiary academic referral centers were used. A total of 13,608 ultrasonography images from 3446 patients who underwent surgical resection, biopsy, or transient elastography were used for training a DCNN for the prediction of the METAVIR score. Pathological specimens or estimated METAVIR scores derived from transient elastography were used as a reference standard. A four-class model (F0 vs. F1 vs. F23 vs. F4) was developed. Diagnostic performance of the algorithm was validated on a separate internal test set of 266 patients with 300 images and external test set of 572 patients with 1232 images. Performance in classification of cirrhosis was compared between the DCNN and five radiologists. Results: The accuracy of the four-class model was 83.5% and 76.4% on the internal and external test set, respectively. The area under the receiver operating characteristic curve (AUC) for classification of cirrhosis (F4) was 0.901 (95% confidence interval [CI], 0.865–0.937) on the internal test set and 0.857 (95% CI, 0.825–0.889) on the external test set, respectively. The AUC of the DCNN for classification of cirrhosis (0.857) was significantly higher than that of all five radiologists (AUC range, 0.656–0.816; p value < 0.05) using the external test set. Conclusions: The DCNN showed high accuracy for determining METAVIR score using ultrasonography images and achieved better performance than that of radiologists in the diagnosis of cirrhosis. Key Points: • DCNN accurately classified the ultrasonography images according to the METAVIR score. • The AUROC of this algorithm for cirrhosis assessment was significantly higher than that of radiologists. • DCNN using US images may offer an alternative tool for monitoring liver fibrosis.

Original languageEnglish
Pages (from-to)1264-1273
Number of pages10
JournalEuropean Radiology
Volume30
Issue number2
DOIs
StatePublished - 1 Feb 2020

Fingerprint

Liver Cirrhosis
Ultrasonography
Fibrosis
Learning
Area Under Curve
Elasticity Imaging Techniques
Confidence Intervals
Tertiary Care Centers
ROC Curve
Biopsy
Radiologists

Keywords

  • Deep learning
  • Fibrosis
  • Liver
  • Ultrasonography

Cite this

Lee, Jeong Hyun ; Joo, Ijin ; Kang, Tae Wook ; Paik, Yong Han ; Sinn, Dong Hyun ; Ha, Sang Yun ; Kim, Kyunga ; Choi, Choonghwan ; Lee, Gunwoo ; Yi, Jonghyon ; Bang, Won Chul. / Deep learning with ultrasonography : automated classification of liver fibrosis using a deep convolutional neural network. In: European Radiology. 2020 ; Vol. 30, No. 2. pp. 1264-1273.
@article{c6b2b3cfa9664e20807ccee8f8398428,
title = "Deep learning with ultrasonography: automated classification of liver fibrosis using a deep convolutional neural network",
abstract = "Objectives: The aim of this study was to develop a deep convolutional neural network (DCNN) for the prediction of the METAVIR score using B-mode ultrasonography images. Methods: Datasets from two tertiary academic referral centers were used. A total of 13,608 ultrasonography images from 3446 patients who underwent surgical resection, biopsy, or transient elastography were used for training a DCNN for the prediction of the METAVIR score. Pathological specimens or estimated METAVIR scores derived from transient elastography were used as a reference standard. A four-class model (F0 vs. F1 vs. F23 vs. F4) was developed. Diagnostic performance of the algorithm was validated on a separate internal test set of 266 patients with 300 images and external test set of 572 patients with 1232 images. Performance in classification of cirrhosis was compared between the DCNN and five radiologists. Results: The accuracy of the four-class model was 83.5{\%} and 76.4{\%} on the internal and external test set, respectively. The area under the receiver operating characteristic curve (AUC) for classification of cirrhosis (F4) was 0.901 (95{\%} confidence interval [CI], 0.865–0.937) on the internal test set and 0.857 (95{\%} CI, 0.825–0.889) on the external test set, respectively. The AUC of the DCNN for classification of cirrhosis (0.857) was significantly higher than that of all five radiologists (AUC range, 0.656–0.816; p value < 0.05) using the external test set. Conclusions: The DCNN showed high accuracy for determining METAVIR score using ultrasonography images and achieved better performance than that of radiologists in the diagnosis of cirrhosis. Key Points: • DCNN accurately classified the ultrasonography images according to the METAVIR score. • The AUROC of this algorithm for cirrhosis assessment was significantly higher than that of radiologists. • DCNN using US images may offer an alternative tool for monitoring liver fibrosis.",
keywords = "Deep learning, Fibrosis, Liver, Ultrasonography",
author = "Lee, {Jeong Hyun} and Ijin Joo and Kang, {Tae Wook} and Paik, {Yong Han} and Sinn, {Dong Hyun} and Ha, {Sang Yun} and Kyunga Kim and Choonghwan Choi and Gunwoo Lee and Jonghyon Yi and Bang, {Won Chul}",
year = "2020",
month = "2",
day = "1",
doi = "10.1007/s00330-019-06407-1",
language = "English",
volume = "30",
pages = "1264--1273",
journal = "European radiology",
issn = "0938-7994",
publisher = "Springer Verlag",
number = "2",

}

Lee, JH, Joo, I, Kang, TW, Paik, YH, Sinn, DH, Ha, SY, Kim, K, Choi, C, Lee, G, Yi, J & Bang, WC 2020, 'Deep learning with ultrasonography: automated classification of liver fibrosis using a deep convolutional neural network', European Radiology, vol. 30, no. 2, pp. 1264-1273. https://doi.org/10.1007/s00330-019-06407-1

Deep learning with ultrasonography : automated classification of liver fibrosis using a deep convolutional neural network. / Lee, Jeong Hyun; Joo, Ijin; Kang, Tae Wook; Paik, Yong Han; Sinn, Dong Hyun; Ha, Sang Yun; Kim, Kyunga; Choi, Choonghwan; Lee, Gunwoo; Yi, Jonghyon; Bang, Won Chul.

In: European Radiology, Vol. 30, No. 2, 01.02.2020, p. 1264-1273.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Deep learning with ultrasonography

T2 - automated classification of liver fibrosis using a deep convolutional neural network

AU - Lee, Jeong Hyun

AU - Joo, Ijin

AU - Kang, Tae Wook

AU - Paik, Yong Han

AU - Sinn, Dong Hyun

AU - Ha, Sang Yun

AU - Kim, Kyunga

AU - Choi, Choonghwan

AU - Lee, Gunwoo

AU - Yi, Jonghyon

AU - Bang, Won Chul

PY - 2020/2/1

Y1 - 2020/2/1

N2 - Objectives: The aim of this study was to develop a deep convolutional neural network (DCNN) for the prediction of the METAVIR score using B-mode ultrasonography images. Methods: Datasets from two tertiary academic referral centers were used. A total of 13,608 ultrasonography images from 3446 patients who underwent surgical resection, biopsy, or transient elastography were used for training a DCNN for the prediction of the METAVIR score. Pathological specimens or estimated METAVIR scores derived from transient elastography were used as a reference standard. A four-class model (F0 vs. F1 vs. F23 vs. F4) was developed. Diagnostic performance of the algorithm was validated on a separate internal test set of 266 patients with 300 images and external test set of 572 patients with 1232 images. Performance in classification of cirrhosis was compared between the DCNN and five radiologists. Results: The accuracy of the four-class model was 83.5% and 76.4% on the internal and external test set, respectively. The area under the receiver operating characteristic curve (AUC) for classification of cirrhosis (F4) was 0.901 (95% confidence interval [CI], 0.865–0.937) on the internal test set and 0.857 (95% CI, 0.825–0.889) on the external test set, respectively. The AUC of the DCNN for classification of cirrhosis (0.857) was significantly higher than that of all five radiologists (AUC range, 0.656–0.816; p value < 0.05) using the external test set. Conclusions: The DCNN showed high accuracy for determining METAVIR score using ultrasonography images and achieved better performance than that of radiologists in the diagnosis of cirrhosis. Key Points: • DCNN accurately classified the ultrasonography images according to the METAVIR score. • The AUROC of this algorithm for cirrhosis assessment was significantly higher than that of radiologists. • DCNN using US images may offer an alternative tool for monitoring liver fibrosis.

AB - Objectives: The aim of this study was to develop a deep convolutional neural network (DCNN) for the prediction of the METAVIR score using B-mode ultrasonography images. Methods: Datasets from two tertiary academic referral centers were used. A total of 13,608 ultrasonography images from 3446 patients who underwent surgical resection, biopsy, or transient elastography were used for training a DCNN for the prediction of the METAVIR score. Pathological specimens or estimated METAVIR scores derived from transient elastography were used as a reference standard. A four-class model (F0 vs. F1 vs. F23 vs. F4) was developed. Diagnostic performance of the algorithm was validated on a separate internal test set of 266 patients with 300 images and external test set of 572 patients with 1232 images. Performance in classification of cirrhosis was compared between the DCNN and five radiologists. Results: The accuracy of the four-class model was 83.5% and 76.4% on the internal and external test set, respectively. The area under the receiver operating characteristic curve (AUC) for classification of cirrhosis (F4) was 0.901 (95% confidence interval [CI], 0.865–0.937) on the internal test set and 0.857 (95% CI, 0.825–0.889) on the external test set, respectively. The AUC of the DCNN for classification of cirrhosis (0.857) was significantly higher than that of all five radiologists (AUC range, 0.656–0.816; p value < 0.05) using the external test set. Conclusions: The DCNN showed high accuracy for determining METAVIR score using ultrasonography images and achieved better performance than that of radiologists in the diagnosis of cirrhosis. Key Points: • DCNN accurately classified the ultrasonography images according to the METAVIR score. • The AUROC of this algorithm for cirrhosis assessment was significantly higher than that of radiologists. • DCNN using US images may offer an alternative tool for monitoring liver fibrosis.

KW - Deep learning

KW - Fibrosis

KW - Liver

KW - Ultrasonography

UR - http://www.scopus.com/inward/record.url?scp=85072122366&partnerID=8YFLogxK

U2 - 10.1007/s00330-019-06407-1

DO - 10.1007/s00330-019-06407-1

M3 - Article

C2 - 31478087

AN - SCOPUS:85072122366

VL - 30

SP - 1264

EP - 1273

JO - European radiology

JF - European radiology

SN - 0938-7994

IS - 2

ER -