Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2023 Jun 8;23(12):5450.
doi: 10.3390/s23125450.

A Hierarchical Siamese Network for Noninvasive Staging of Liver Fibrosis Based on US Image Pairs of the Liver and Spleen

Xue Wang  1 , Ling Song  2 , Yan Zhuang  1 , Lin Han  1 , Ke Chen  1 , Jiangli Lin  1 , Yan Luo  2
Affiliations

A Hierarchical Siamese Network for Noninvasive Staging of Liver Fibrosis Based on US Image Pairs of the Liver and Spleen

Xue Wang et al. Sensors (Basel). .

Abstract

Due to the heterogeneity of ultrasound (US) images and the indeterminate US texture of liver fibrosis (LF), automatic evaluation of LF based on US images is still challenging. Thus, this study aimed to propose a hierarchical Siamese network that combines the information from liver and spleen US images to improve the accuracy of LF grading. There were two stages in the proposed method. In stage one, a dual-channel Siamese network was trained to extract features from paired liver and spleen patches that were cropped from US images to avoid vascular interferences. Subsequently, the L1 distance was used to quantify the liver-spleen differences (LSDs). In stage two, the pretrained weights from stage one were transferred into the Siamese feature extractor of the LF staging model, and a classifier was trained using the fusion of the liver and LSD features for LF staging. This study was retrospectively conducted on US images of 286 patients with histologically proven liver fibrosis stages. Our method achieved a precision and sensitivity of 93.92% and 91.65%, respectively, for cirrhosis (S4) diagnosis, which is about 8% higher than that of the baseline model. The accuracy of the advanced fibrosis (≥S3) diagnosis and the multi-staging of fibrosis (≤S2 vs. S3 vs. S4) both improved about 5% to reach 90.40% and 83.93%, respectively. This study proposed a novel method that combined hepatic and splenic US images and improved the accuracy of LF staging, which indicates the great potential of liver-spleen texture comparison in noninvasive assessment of LF based on US images.

Keywords: Siamese network; US images; liver fibrosis; liver–spleen texture comparison.

PubMed Disclaimer

Conflict of interest statement

The authors declare no conflict of interest.

Figures

Figure 1
Figure 1
US images with different degrees of LF (S1–S4 means liver fibrosis stages 1–4, which are determined through liver biopsy by pathologists).
Figure 2
Figure 2
US image signs of fibrotic liver and spleen (S1–S4 means liver fibrosis stages 1–4, which are determined through liver biopsy by pathologists. The red arrows pointing to US signs of fibrosis as examples). (a) S1: Rough echo of liver parenchyma. (b) S2: “strip pattern” liver fibrosis. (c) S3: significantly rough echo of liver parenchyma with uneven distribution. (d) S4: hyperplasia nodules of liver parenchyma.
Figure 2
Figure 2
US image signs of fibrotic liver and spleen (S1–S4 means liver fibrosis stages 1–4, which are determined through liver biopsy by pathologists. The red arrows pointing to US signs of fibrosis as examples). (a) S1: Rough echo of liver parenchyma. (b) S2: “strip pattern” liver fibrosis. (c) S3: significantly rough echo of liver parenchyma with uneven distribution. (d) S4: hyperplasia nodules of liver parenchyma.
Figure 3
Figure 3
The workflow of image processing.
Figure 4
Figure 4
Overall algorithm scheme. Subfigure (a) shows the process of data preparation. Subfigure (b) shows the dual-channel feature extractor for liver and spleen feature extraction and the KDE method for LSD analysis. Subfigure (c) shows the dual-channel LF staging model. Abbreviations: ROI = region of interest, CNN = convolutional neural network, KDE = kernel density estimation, WT = weight transfer, LSD = liver–spleen difference, FC = fully connected, LF = liver fibrosis, FF = feature fusion.
Figure 5
Figure 5
Structure of dual-channel CNN feature extractor.
Figure 6
Figure 6
Feature fusion for FC classifier training.
Figure 7
Figure 7
Kernel density estimation (KDE, left) and cumulative distribution function (CDF, right). (a) Noncirrhotic vs. cirrhotic. (b) Mild fibrosis vs. advanced fibrosis.
Figure 7
Figure 7
Kernel density estimation (KDE, left) and cumulative distribution function (CDF, right). (a) Noncirrhotic vs. cirrhotic. (b) Mild fibrosis vs. advanced fibrosis.
Figure 8
Figure 8
Examples of LSD values measured in cases with different degrees of fibrosis (S1–S4 means liver fibrosis stages 1–4, which were determined though liver biopsy by pathologists). (a) S0/S1. (b) S2. (c) S3. (d) S4.
Figure 9
Figure 9
Confusion matrix of cirrhosis diagnosis based on dual-channel AlexNet.
Figure 10
Figure 10
Confusion matrix of advanced fibrosis diagnosis based on dual-channel AlexNet.
Figure 11
Figure 11
Confusion matrix of LF multi-staging based on dual-channel AlexNet.
See this image and copyright information in PMC

Similar articles

See all similar articles

References

    1. Bataller R., Brenner D.A. Liver fibrosis. J. Clin. Investig. 2005;115:209–218. doi: 10.1172/JCI24282. - DOI - PMC - PubMed
    1. Zhang C.-Y., Yuan W.-G., He P., Lei J.-H., Wang C.-X. Liver fibrosis and hepatic stellate cells: Etiology, pathological hallmarks and therapeutic targets. World J. Gastroenterol. 2016;22:10512–10522. doi: 10.3748/wjg.v22.i48.10512. - DOI - PMC - PubMed
    1. Aydin M.M., Akcali K.C. Liver fibrosis. Turk. J. Gastroenterol. 2018;29:14–21. doi: 10.5152/tjg.2018.17330. - DOI - PMC - PubMed
    1. Lai M., Afdhal N.H. Liver Fibrosis Determination. Gastroenterol. Clin. N. Am. 2019;48:281–289. doi: 10.1016/j.gtc.2019.02.002. - DOI - PubMed
    1. Bravo A.A., Sheth S.G., Chopra S. Current concepts: Liver biopsy. N. Engl. J. Med. 2001;344:495–500. doi: 10.1056/NEJM200102153440706. - DOI - PubMed

LinkOut - more resources