Deep learning for staging liver fibrosis on CT: a pilot study
- 681 Downloads
To investigate whether liver fibrosis can be staged by deep learning techniques based on CT images.
This clinical retrospective study, approved by our institutional review board, included 496 CT examinations of 286 patients who underwent dynamic contrast-enhanced CT for evaluations of the liver and for whom histopathological information regarding liver fibrosis stage was available. The 396 portal phase images with age and sex data of patients (F0/F1/F2/F3/F4 = 113/36/56/66/125) were used for training a deep convolutional neural network (DCNN); the data for the other 100 (F0/F1/F2/F3/F4 = 29/9/14/16/32) were utilised for testing the trained network, with the histopathological fibrosis stage used as reference. To improve robustness, additional images for training data were generated by rotating or parallel shifting the images, or adding Gaussian noise. Supervised training was used to minimise the difference between the liver fibrosis stage and the fibrosis score obtained from deep learning based on CT images (FDLCT score) output by the model. Testing data were input into the trained DCNNs to evaluate their performance.
The FDLCT scores showed a significant correlation with liver fibrosis stage (Spearman's correlation coefficient = 0.48, p < 0.001). The areas under the receiver operating characteristic curves (with 95% confidence intervals) for diagnosing significant fibrosis (≥ F2), advanced fibrosis (≥ F3) and cirrhosis (F4) by using FDLCT scores were 0.74 (0.64–0.85), 0.76 (0.66–0.85) and 0.73 (0.62–0.84), respectively.
Liver fibrosis can be staged by using a deep learning model based on CT images, with moderate performance.
• Liver fibrosis can be staged by a deep learning model based on magnified CT images including the liver surface, with moderate performance.
• Scores from a trained deep learning model showed moderate correlation with histopathological liver fibrosis staging.
• Further improvement are necessary before utilisation in clinical settings.
KeywordsLiver cirrhosis Artificial intelligence Multidetector computed tomography ROC curve
Area under the receiver operating characteristic curve
Deep convolutional neural network
Digital Imaging and Communications in Medicine
Fibrosis score obtained from deep learning based on CT images
Joint Photographic Experts Group
Magnetic resonance elastography
Receiver operating characteristic
Region of interest
The authors state that this work has not received any funding.
Compliance with ethical standards
The scientific guarantor of this publication is Koichiro Yasaka.
Conflict of interest
The authors of this manuscript declare no relationships with any companies whose products or services may be related to the subject matter of the article.
Statistics and biometry
No complex statistical methods were necessary for this paper.
Written informed consent was waived by the institutional review board.
Institutional review board approval was obtained.
• diagnostic or prognostic study
• performed at one institution
- 10.Krizhevsky A, Sutskever I, Hinton G (2012) ImageNet classification with deep convolutional neural networks. Advances in Neural Information Processing System 25 (NIPS 2012). https://papersnipscc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks. Accessed 20 Jan 2018
- 11.Szegedy C, Liu W, Jia Y et al (2014) Going deeper with convolutions. Cornell University Library. https://arxivorg/abs/14094842. Accessed 20 Jan 2018
- 12.He K, Zhang X, Ren S, Sun J (2015) Deep residual learning for image recognition. Cornell University Library. https://arxivorg/abs/151203385. Accessed 20 Jan 2018
- 13.Andrearczyk V, Whelan PF (2016) Using filter banks in convolutional neural networks for texture classification. Cornell University Library. https://arxiv.org/abs/1601.02919. Accessed 20 Jan 2018
- 18.Leynes AP, Yang J, Wiesinger F et al (2017) Direct pseudoCT generation for pelvis PET/MRI attenuation correction using deep convolutional neural networks with multi-parametric MRI: zero echo-time and Dixon deep pseudoCT (ZeDD-CT). J Nucl Med. https://doi.org/10.2967/jnumed.117.198051 CrossRefGoogle Scholar
- 25.Nair V, Hinton G (2010) Rectified linear units improve restricted Boltzmann machines. International conference on machine learning. http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.165.6419&rank=1. Accessed 20 Jan 2018
- 26.Ioffe S, Szegedy C (2015) Batch normalization: accelerating deep network training by reducing internal covariate shift. Cornell University Library. http://arxiv.org/abs/1502.03167. Accessed 20 Jan 2018
- 28.Duchi J, Hazan E, Singer Y (2011) Adaptive subgradient methods for online learning and stochastic optimization. J Mach Learn Res 12:2121–2159Google Scholar
- 31.Singh S, Venkatesh SK, Wang Z et al (2015) Diagnostic performance of magnetic resonance elastography in staging liver fibrosis: a systematic review and meta-analysis of individual participant data. Clin Gastroenterol Hepatol 13(440-451):e446Google Scholar