A 3D Deep Neural Network for Liver Volumetry in 3T Contrast-Enhanced MRI

Abstract

Purpose To create a fully automated, reliable, and fast segmentation tool for Gd-EOB-DTPA-enhanced MRI scans using deep learning.

Materials and Methods Datasets of Gd-EOB-DTPA-enhanced liver MR images of 100 patients were assembled. Ground truth segmentation of the hepatobiliary phase images was performed manually. Automatic image segmentation was achieved with a deep convolutional neural network.

Results Our neural network achieves an intraclass correlation coefficient (ICC) of 0.987, a Sørensen–Dice coefficient of 96.7 ± 1.9 % (mean ± std), an overlap of 92 ± 3.5 %, and a Hausdorff distance of 24.9 ± 14.7 mm compared with two expert readers who corresponded to an ICC of 0.973, a Sørensen–Dice coefficient of 95.2 ± 2.8 %, and an overlap of 90.9 ± 4.9 %. A second human reader achieved a Sørensen–Dice coefficient of 95 % on a subset of the test set.

Conclusion Our study introduces a fully automated liver volumetry scheme for Gd-EOB-DTPA-enhanced MR imaging. The neural network achieves competitive concordance with the ground truth regarding ICC, Sørensen–Dice, and overlap compared with manual segmentation. The neural network performs the task in just 60 seconds.

Key Points: 

• The proposed neural network helps to segment the liver accurately, providing detailed information about patient-specific liver anatomy and volume.

• With the help of a deep learning-based neural network, fully automatic segmentation of the liver on MRI scans can be performed in seconds.

• A fully automatic segmentation scheme makes liver segmentation on MRI a valuable tool for treatment planning.

Citation Format

• Winther H, Hundt C, Ringe KI et al. A 3D Deep Neural Network for Liver Volumetry in 3T Contrast-Enhanced MRI. Fortschr Röntgenstr 2021; 193: 305 – 314

Zusammenfassung

Ziel Ziel dieser Studie war es, eine vollautomatische und zuverlässige Lebervolumetrie in der kontrastverstärkten MRT basierend auf 3D-Deep-Learning-Algorithmen zu entwickeln.

Material und Methoden Datensätze von Gd-EOB-DTPA-verstärkten Leber-MR-Bildern von 100 Patienten wurden von einem in der hepatobiliären Bildgebung erfahrenen Radiologen manuell segmentiert und als Grundwahrheitssegmentierung angenommen. Die Datensätze wurden mittels einem Kreuzvalidierungsverfahren (k = 4) in Trainings- und Validierungsdatensatz eingeteilt und einem neuronalen Netzwerk zur automatischen Bildsegmentierung zugeführt. Zusätzlich wurde ein Teil der Daten (n = 9) von einem zweiten Radiologen zur Bestimmung einer Interobserver Variability segmentiert.

Ergebnisse Die manuelle Segmentierung erreichte einen Inter-Klassen-Korrelationskoeffizienten (ICC) von 0,973, einen Sørensen-Dice-Index von 95,2 ± 2,8 % und eine Überlappung von 90,9 ± 4,9 %. Das neuronale Netzwerk erreichte einen ICC von 0,98, einen Sørensen-Dice-Index von 96 ± 1,9 % und eine Überlappung von 92 ± 3,5 % sowie eine Hausdorff-Distanz von 24,9 ± 14,7 mm.

Schlussfolgerung Diese Studie präsentiert ein vollautomatisches Lebervolumetrie-Schema für MR-Bildgebung. Das neuronale Netzwerk erreichte eine kompetitive Übereinstimmung mit der Grundwahrheit bezüglich ICC, Sørensen-Dice-Index und Überlappung im Vergleich zu einer manuellen Segmentierung. Das neuronale Netzwerk erledigte die Aufgabe in nur 60 Sekunden.

Kernaussagen: 

• Das vorgeschlagene neuronale Netzwerk hilft bei der genauen Segmentierung der Leber und liefert detaillierte Informationen über die patientenspezifische Anatomie und das Volumen der Leber.

• Mithilfe eines neuronalen Netzes kann eine vollautomatische Segmentierung der Leber in MRT-Scans in Sekundenschnelle durchgeführt werden.

• Ein vollautomatisches Segmentierungsschema macht die Lebersegmentierung in der MRT zu einem wertvollen Instrument für die Behandlungsplanung.

Key words

liver segmentation - liver volumetry - semantic segmentation - fully automated segmentation - contrast-enhanced liver MRI

Publication History

Received: 10 February 2020

Accepted: 03 August 2020

Article published online:
03 September 2020

• References

• 1 Ribero D, Amisano M, Bertuzzo F. et al Measured versus estimated total liver volume to preoperatively assess the adequacy of the future liver remnant: which method should we use?. Ann Surg 2013; 258: 801-806 ; discussion 806–807
• 2 Yigitler C, Farges O, Kianmanesh R. et al. The small remnant liver after major liver resection: how common and how relevant?. Liver transplantation: official publication of the American Association for the Study of Liver Diseases and the International Liver Transplantation Society 2003; 9: S18-S25
• 3 Guglielmi A, Ruzzenente A, Conci S. et al. How much remnant is enough in liver resection?. Digestive surgery 2012; 29: 6-17
• 4 Heimann T, van Ginneken B, Styner MA. et al. Comparison and evaluation of methods for liver segmentation from CT datasets. IEEE transactions on medical imaging 2009; 28: 1251-1265
• 5 Li G, Chen X, Shi F. et al. Automatic Liver Segmentation Based on Shape Constraints and Deformable Graph Cut in CT Images. IEEE transactions on image processing: a publication of the IEEE Signal Processing Society 2015; 24: 5315-5329
• 6 Suzuki K, Epstein ML, Kohlbrenner R. et al. Quantitative radiology: automated CT liver volumetry compared with interactive volumetry and manual volumetry. Am J Roentgenol American journal of roentgenology 2011; 197: W706-W712
• 7 Kumar SS, Moni RS, Rajeesh J. Automatic liver and lesion segmentation: a primary step in diagnosis of liver diseases. Signal Image Video Process 2013; 7: 163-172
• 8 Goryawala M, Gulec S, Bhatt R. et al. A low-interaction automatic 3D liver segmentation method using computed tomography for selective internal radiation therapy. Biomed Res Int 2014; 2014: 198015
• 9 Yuqian Z, Yunlong Z, Xiaofang W. et al Fuzzy C-means clustering-based multilayer perceptron neural network for liver CT images automatic segmentation. In, 2010 Chinese control and decision conference Xuzhou, China: IEEE; 2010: 3423-3427
• 10 Peng J, Hu P, Lu F. et al. 3D liver segmentation using multiple region appearances and graph cuts. Med Phys 2015; 42: 6840-6852
• 11 Priyadarsini S, Selvathi D. Survey on segmentation of liver from CT images. In, 2012 IEEE International conference on advanced communication control and computing technologies (ICACCCT) Ramanathapuram, India: IEEE; 2012: 234-238
• 12 Gotra A, Sivakumaran L, Chartrand G. et al. Liver segmentation: indications, techniques and future directions. Insights into imaging 2017; 8: 377-392
• 13 Huynh HT, Karademir I, Oto A. et al. Computerized liver volumetry on MRI by using 3D geodesic active contour segmentation. Am J Roentgenol American journal of roentgenology 2014; 202: 152-159
• 14 Huynh HT, Le-Trong N, Bao PT. et al. Fully automated MR liver volumetry using watershed segmentation coupled with active contouring. Int J Comput Assist Radiol Surg 2017; 12: 235-243
• 15 Nassif A, Jia J, Keiser M. et al. Visualization of hepatic uptake transporter function in healthy subjects by using gadoxetic acid-enhanced MR imaging. Radiology 2012; 264: 741-750
• 16 Kobayashi S, Matsui O, Gabata T. et al. Relationship between signal intensity on hepatobiliary phase of gadolinium ethoxybenzyl diethylenetriaminepentaacetic acid (Gd-EOB-DTPA)-enhanced MR imaging and prognosis of borderline lesions of hepatocellular carcinoma. European journal of radiology 2012; 81: 3002-3009
• 17 Van Beers BE, Pastor CM, Hussain HK. Primovist, Eovist: what to expect?. J Hepatol 2012; 57: 421-429
• 18 Bluemke DA, Sahani D, Amendola M. et al. Efficacy and safety of MR imaging with liver-specific contrast agent: U.S. multicenter phase III study. Radiology 2005; 237: 89-98
• 19 van Montfoort JE, Stieger B, Meijer DK. et al. Hepatic uptake of the magnetic resonance imaging contrast agent gadoxetate by the organic anion transporting polypeptide Oatp1. The Journal of pharmacology and experimental therapeutics 1999; 290: 153-157
• 20 Weinmann HJ, Bauer H, Frenzel T. et al. Mechanism of hepatic uptake of gadoxetate disodium. Academic radiology 1996; 3: S232-S234
• 21 Schuhmann-Giampieri G, Schmitt-Willich H, Frenzel T. et al. Biliary excretion and pharmacokinetics of a gadolinium chelate used as a liver-specific contrast agent for magnetic resonance imaging in the rat. Journal of pharmaceutical sciences 1993; 82: 799-803
• 22 Weinmann HJ, Schuhmann-Giampieri G, Schmitt-Willich H. et al A new lipophilic gadolinium chelate as a tissue-specific contrast medium for MRI. Magnetic resonance in medicine: official journal of the Society of Magnetic Resonance in Medicine/Society of Magnetic Resonance in Medicine 1991; 22: 233-237 ; discussion 242
• 23 Clement O, Muhler A, Vexler V. et al. Gadolinium-ethoxybenzyl-DTPA, a new liver-specific magnetic resonance contrast agent. Kinetic and enhancement patterns in normal and cholestatic rats. Investigative radiology 1992; 27: 612-619
• 24 Rosset A, Spadola L, Ratib O. OsiriX: an open-source software for navigating in multidimensional DICOM images. J Digit Imaging 2004; 17: 205-216
• 25 Yushkevich PA, Piven J, Hazlett HC. et al. User-guided 3D active contour segmentation of anatomical structures: significantly improved efficiency and reliability. NeuroImage 2006; 31: 1116-1128
• 26 Çiçek Ö, Abdulkadir A, Lienkamp SS. et al. 3D U-net: learning dense volumetric segmentation from sparse annotation. In: Ourselin S, Joskowicz L, Sabuncu MR. et al. eds Medical image computing and computer-assisted intervention – MICCAI 2016. Cham: Springer International Publishing; 2016: 424-432
• 27 Winther HB, Hundt C, Schmidt B. et al. ν-net: deep learning for generalized biventricular mass and function parameters using multicenter cardiac MRI data. JACC Cardiovasc Imaging 2018; 11: 1036-1038
• 28 Glorot X, Bengio Y. Understanding the difficulty of training deep feed-forward neural networks. In: Yee Whye T, Mike T. eds Proceedings of the thirteenth International conference on artificial intelligence and statistics. PMLR; 2010: 249-256
• 29 Kingma DP, Ba J. Adam: a method for stochastic optimization. ArXiv e-prints; 2014
• 30 R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2013
• 31 Freiman M, Eliassaf O, Taieb Y. et al. An iterative Bayesian approach for nearly automatic liver segmentation: algorithm and validation. Int J Comput Assist Radiol Surg 2008; 3: 439
• 32 Gao L, Heath DG, Kuszyk BS. et al. Automatic liver segmentation technique for three-dimensional visualization of CT data. Radiology 1996; 201: 359-364
• 33 Seo KS. Improved fully automatic liver segmentation using histogram tail threshold algorithms. In: Sunderam VS, van Albada GD, Sloot PMA. et al. eds, Computational science – ICCS 2005. Berlin, Heidelberg: Springer Berlin Heidelberg; 2005: 822-825
• 34 Hermoye L, Laamari-Azjal I, Cao Z. et al. Liver segmentation in living liver transplant donors: comparison of semiautomatic and manual methods. Radiology 2005; 234: 171-178
• 35 Chartrand G, Cresson T, Chav R. et al. Liver segmentation on CT and MR using laplacian mesh optimization. IEEE Trans Biomed Eng 2017; 64: 2110-2121
• 36 Cheng K, Gu L, Xu J. A novel shape prior based level set method for liver segmentation from MR images. In, 2008 International conference on information technology and applications in biomedicine Shenzhen, China: IEEE; 2008: 144-147
• 37 Chan TF, Vese LA. Active contours without edges. IEEE transactions on image processing: a publication of the IEEE Signal Processing Society 2001; 10: 266-277
• 38 Gloger O, Kühn J, Stanski A. et al. A fully automatic three-step liver segmentation method on LDA-based probability maps for multiple contrast MR images. Magnetic resonance imaging 2010; 28: 882-897
• 39 Chartrand G, Belanger C, Gotra A. et al Accuracy of a semi-automated liver segmentation method using CT scan. In, European Congress of Radiology Vienna, Austria: ESR; 2013: 1-20
• 40 Yan Z, Zhang S, Tan C. et al. Atlas-based liver segmentation and hepatic fat-fraction assessment for clinical trials. Comput Med Imaging Graph 2015; 41: 80-92
• 41 Hu P, Wu F, Peng J. et al. Automatic 3D liver segmentation based on deep learning and globally optimized surface evolution. Phys Med Biol 2016; 61: 8676-8698
• 42 Lu F, Wu F, Hu P. et al. Automatic 3D liver location and segmentation via convolutional neural network and graph cut. Int J Comput Assist Radiol Surg 2017; 12: 171-182
• 43 Dror IE. Perception is far from perfection: the role of the brain and mind in constructing realities. Behav Brain Sci 2005; 28: 763-763
• 44 Dror IE, Busemeyer JR, Basola B. Decision making under time pressure: an independent test of sequential sampling models. Memory & cognition 1999; 27: 713-725
• 45 Verloh N, Haimerl M, Rennert J. et al. Impact of liver cirrhosis on liver enhancement at Gd-EOB-DTPA enhanced MRI at 3 Tesla. European journal of radiology 2013; 82: 1710-1715
• 46 Verloh N, Haimerl M, Zeman F. et al. Assessing liver function by liver enhancement during the hepatobiliary phase with Gd-EOB-DTPA-enhanced MRI at 3 Tesla. Eur Radiol 2014; 24: 1013-1019
• 47 Verloh N, Utpatel K, Haimerl M. et al. Liver fibrosis and Gd-EOB-DTPA-enhanced MRI: A histopathologic correlation. Scientific reports 2015; 5: 15408
• 48 López-Mir F, Naranjo V, Angulo J. et al. Liver segmentation in MRI: a fully automatic method based on stochastic partitions. Comput Methods Programs Biomed 2014; 114: 11-28
• 49 Bereciartu A, Picon A, Galdran A. et al. Automatic 3D model-based method for liver segmentation in MRI based on active contours and total variation minimization. Biomed Signal Process Control 2015; 20: 71-77