Skip to main content


Log in

Segmentation of the aorta in systolic phase from 4D flow MRI: multi-atlas vs. deep learning

  • Research Article
  • Published:
Magnetic Resonance Materials in Physics, Biology and Medicine Aims and scope Submit manuscript



In the management of the aortic aneurysm, 4D flow magnetic resonance Imaging provides valuable information for the computation of new biomarkers using computational fluid dynamics (CFD). However, accurate segmentation of the aorta is required. Thus, our objective is to evaluate the performance of two automatic segmentation methods on the calculation of aortic wall pressure.


Automatic segmentation of the aorta was performed with methods based on deep learning and multi-atlas using the systolic phase in the 4D flow MRI magnitude image of 36 patients. Using mesh morphing, isotopological meshes were generated, and CFD was performed to calculate the aortic wall pressure. Node-to-node comparisons of the pressure results were made to identify the most robust automatic method respect to the pressures obtained with a manually segmented model.


Deep learning approach presented the best segmentation performance with a mean Dice similarity coefficient and a mean Hausdorff distance (HD) equal to 0.92+/− 0.02 and 21.02+/− 24.20 mm, respectively. At the global level HD is affected by the performance in the abdominal aorta. Locally, this distance decreases to 9.41+/− 3.45 and 5.82+/− 6.23 for the ascending and descending thoracic aorta, respectively. Moreover, with respect to the pressures from the manual segmentations, the differences in the pressures computed from deep learning were lower than those computed from multi-atlas method.


To reduce biases in the calculation of aortic wall pressure, accurate segmentation is needed, particularly in regions with high blood flow velocities. Thus, the deep learning segmen-tation method should be preferred.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9

Similar content being viewed by others


  1. Pinard A, Jones GT, Milewicz DM (2019) Genetics of thoracic and abdominal aortic diseases: aneurysms, dissections, and ruptures. Circ Res 124(4):588–606

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  2. Kuzmik GA, Sang AX, Elefteriades JA (2012) Natural history of thoracic aortic aneurysms. J Vasc Surg 56(2):565–571

    Article  PubMed  Google Scholar 

  3. Ehrman JK, Fernandez AB, Myers J, Oh P, Thompson PD, Keteyian SJ (2020) Aortic aneurysm: diagnosis, management, exercise testing, and training. J Cardiopulm Rehabil Prev 40(4):215–223

    Article  PubMed  Google Scholar 

  4. Adamo L, Braverman AC (2015) Surgical threshold for bicuspid aortic valve aneurysm: a case for individual decision-making. Heart 101(17):1361–1367

    Article  CAS  PubMed  Google Scholar 

  5. Pape L, Tsai T, Isselbacher E, Oh J, O’gara P, Evangelista A, Fattori R, Meinhardt G, Trimarchi S, Bossone E et al (2007) International registry of acute aortic dissection (IRAD) investigators: aortic diameter \(>\) or = 5.5 cm is not a good predictor of type a aortic dissection: observations from the international registry of acute aortic dissection (IRAD). Circulation 116:1120–1127

    Article  PubMed  Google Scholar 

  6. Condemi F, Campisi S, Viallon M, Croisille P, Avril S (2019) Relationship between ascending thoracic aortic aneurysms hemodynamics and biomechanical properties. IEEE Trans Biomed Eng 67(4):949–956

    Article  PubMed  Google Scholar 

  7. Stankovic Z, Allen BD, Garcia J, Jarvis KB, Markl M (2014) 4d flow imaging with MRI. Cardiovasc Diagn Therapy 4(2):173

    Google Scholar 

  8. Ha H, Kim GB, Kweon J, Lee SJ, Kim Y-H, Lee DH, Yang DH, Kim N (2016) Hemodynamic measurement using four-dimensional phase-contrast MRI: quantification of hemodynamic parameters and clinical applications. Korean J Radiol 17(4):445–462

    Article  PubMed  PubMed Central  Google Scholar 

  9. van Pelt R, Nguyen H, ter Haar Romeny B, Vilanova A (2012) Automated segmentation of blood-flow regions in large thoracic arteries using 3D-cine PC-MRI measurements. Int J Comput Assist Radiol Surg 7(2):217–224

    Article  PubMed  Google Scholar 

  10. Berhane H, Scott M, Elbaz M, Jarvis K, McCarthy P, Carr J, Malaisrie C, Avery R, Barker AJ, Robinson JD et al (2020) Fully automated 3D aortic segmentation of 4D flow MRI for hemodynamic analysis using deep learning. Magn Reson Med 84(4):2204–2218

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  11. Fujiwara T, Berhane H, Scott MB, Englund EK, Schäfer M, Fonseca B, Berthusen A, Robinson JD, Rigsby CK, Browne LP et al (2022) Segmentation of the aorta and pulmonary arteries based on 4D flow MRI in the pediatric setting using fully automated multi-site, multi-vendor, and multi-label dense u-net. J Magn Reson Imaging 55(6):1666–1680

    Article  PubMed  Google Scholar 

  12. Bustamante M, Gupta V, Forsberg D, Carlhäll C-J, Engvall J, Ebbers T (2018) Automated multi-atlas segmentation of cardiac 4D flow MRI. Med Image Anal 49:128–140

    Article  PubMed  Google Scholar 

  13. Bustamante M, Viola F, Engvall J, Carlhäll C-J, Ebbers T (2023) Automatic time-resolved cardiovascular segmentation of 4D flow MRI using deep learning. J Magn Reson Imaging 57(1):191–203

    Article  PubMed  Google Scholar 

  14. Leuprecht A, Kozerke S, Boesiger P, Perktold K (2003) Blood flow in the human ascending aorta: a combined MRI and CFD study. J Eng Math 47(3):387–404

    Article  Google Scholar 

  15. Numata S, Itatani K, Kanda K, Doi K, Yamazaki S, Morimoto K, Manabe K, Ikemoto YHK (2016) Blood flow analysis of the aortic arch using computational fluid dynamics. Eur J Cardio-Thorac Surg 49(6):1578–1585

    Article  Google Scholar 

  16. Gülan U, Calen C, Duru F, Holzner M (2018) Blood flow patterns and pressure loss in the ascending aorta: a comparative study on physiological and aneurysmal conditions. J Biomech 76:152–159

    Article  PubMed  Google Scholar 

  17. Renner J, Nadali Najafabadi H, Modin D, Länne T, Karlsson M (2012) Subject-specific aortic wall shear stress estimations using semi-automatic segmentation. Clin Physiol Funct Imaging 32(6):481–491

    Article  CAS  PubMed  Google Scholar 

  18. Alexa M (2002) Recent advances in mesh morphing. Comput Graph Forum 21(2):173–198

    Article  Google Scholar 

  19. Geronzi L, Gasparotti E, Capellini K, Cella U, Groth C, Porziani S, Chiappa A, Celi S, Biancolini ME (2021) High fidelity fluid-structure interaction by radial basis functions mesh adaption of moving walls: a workflow applied to an aortic valve. J Comput Sci 51:101327

    Article  Google Scholar 

  20. Biancolini ME (2012) Mesh morphing and smoothing by means of radial basis functions (RBF): a practical example using fluent and RBF morph. In: Handbook of research on computational science and engineering: theory and practice. IGI Global, pp 347–380

  21. Maquart T, Wenfeng Y, Elguedj T, Gravouil A, Rochette M (2020) 3D volumetric isotopological meshing for finite element and isogeometric based reduced order modeling. Comput Methods Appl Mech Eng 362:112809

    Article  Google Scholar 

  22. Biancolini ME, Ponzini R, Antiga L, Morbiducci U (2012) A new workflow for patient specific image-based hemodynamics: parametric study of the carotid bifurcation. Model objects represented images III. Fundam Methods Appl Comput

  23. Capellini K, Vignali E, Costa E, Gasparotti E, Biancolini ME, Landini L, Positano V, Celi S (2018) Computational fluid dynamic study for ATAA hemodynamics: an integrated image-based and radial basis functions mesh morphing approach. J Biomech Eng 140(11):111007

    Article  PubMed  Google Scholar 

  24. Capellini K, Gasparotti E, Cella U, Costa E, Fanni BM, Groth C, Porziani S, ME B, Celi S, (2021) A novel formulation for the study of the ascending aortic fluid dynamics with in vivo data. Med Eng Phys 91:68–78

  25. Groth C, Porziani S, Biancolini ME, Costa E, Celi S, Capellini K, Rochette M, Morgenthaler V (2018) The medical digital twin assisted by reduced order models and mesh morphing 10

  26. Indrakusuma R, Jalalzadeh H, Planken R, Marquering H, Legemate D, Koelemay M, Balm R (2016) Biomechanical imaging markers as predictors of abdominal aortic aneurysm growth or rupture: a systematic review. Eur J Vasc Endovasc Surg 52(4):475–486

    Article  CAS  PubMed  Google Scholar 

  27. Catalano C, Agnese V, Gentile G, Raffa GM, Pilato M, Pasta S (2021) Atlas-based evaluation of hemodynamic in ascending thoracic aortic aneurysms. Appl Sci 12(1):394

    Article  Google Scholar 

  28. Pasta S, Gentile G, Raffa G, Bellavia D, Chiarello G, Liotta R, Luca A, Scardulla C, Pilato M (2017) In silico shear and intramural stresses are linked to aortic valve morphology in dilated ascending aorta. Eur J Vasc Endovasc Surg 54(2):254–263

    Article  CAS  PubMed  Google Scholar 

  29. Iglesias JE, Sabuncu MR (2015) Multi-atlas segmentation of biomedical images: a survey. Med Image Anal 24(1):205–219

    Article  PubMed  PubMed Central  Google Scholar 

  30. Wang Z, Bovik AC, Sheikh HR, Simoncelli EP (2004) Image quality assessment: from error visibility to structural similarity. IEEE Trans Image Process 13(4):600–612

    Article  PubMed  Google Scholar 

  31. Marin-Castrillon DM, Boucher A, Lin S, Bernard C, Morgant M-C, Cochet A, Lalande A, Bouchot O, Presles B (2021) Multi-atlas segmentation of the aorta from 4D flow MRI: comparison of several fusion strategies. In: International workshop on statistical atlases and computational models of the heart. Springer, pp 3–11

  32. Artaechevarria X, Munoz-Barrutia A, Ortiz-de-Solorzano C (2009) Combination strategies in multi-atlas image segmentation: application to brain MR data. IEEE Trans Med Imaging 28(8):1266–1277

    Article  PubMed  Google Scholar 

  33. Ronneberger O, Fischer P, Brox T (2015) U-net: Convolutional networks for biomedical image segmentation. In: International conference on medical image computing and computer-assisted intervention. Springer, pp 234–241

  34. Çiçek Ö, Abdulkadir A, Lienkamp SS, Brox T, Ronneberger O (2016) 3D U-net: learning dense volumetric segmentation from sparse annotation. In: International conference on medical image computing and computer-assisted intervention. Springer, pp 424–432

  35. Badrinarayanan V, Kendall A, Cipolla R (2017) Segnet: a deep convolutional encoder-decoder architecture for image segmentation. IEEE Trans Pattern Anal Mach Intell 39(12):2481–2495

    Article  PubMed  Google Scholar 

  36. Janssens R, Zeng G, Zheng G (2018) Fully automatic segmentation of lumbar vertebrae from CT images using cascaded 3D fully convolutional networks. In: 2018 IEEE 15th international symposium on biomedical imaging (ISBI 2018). IEEE, pp 893–897

  37. Grassi L, Hraiech N, Schileo E, Ansaloni M, Rochette M, Viceconti M (2011) Evaluation of the generality and accuracy of a new mesh morphing procedure for the human femur. Med Eng Phys 33(1):112–120

    Article  PubMed  Google Scholar 

  38. Buhmann MD (2000) Radial basis functions. Acta Numer 9:1–38

    Article  Google Scholar 

  39. Braaten M, Shyy W (1986) A study of recirculating flow computation using body-fitted coordinates: consistency aspects and mesh skewness. Numer Heat Transf Part A Appl 9(5):559–574

    Google Scholar 

  40. Shore A (2000) Capillaroscopy and the measurement of capillary pressure. Br J Clin Pharmacol 50(6):501–513

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  41. Liang L, Mao W, Sun W (2020) A feasibility study of deep learning for predicting hemodynamics of human thoracic aorta. J Biomech 99:109544

    Article  PubMed  Google Scholar 

  42. Warfield SK, Zou KH, Wells WM (2004) Simultaneous truth and performance level estimation (staple): an algorithm for the validation of image segmentation. IEEE Trans Med Imaging 23(7):903–921

    Article  PubMed  PubMed Central  Google Scholar 

  43. Lalande A, Garreau M, Frouin F (2015) Multi-modality cardiac imaging: processing and analysis. Wiley, New York, pp 184–192 (Chap. 5)

    Google Scholar 

  44. Klein S, Van Der Heide UA, Lips IM, Van Vulpen M, Staring M, Pluim JP (2008) Automatic segmentation of the prostate in 3D MR images by atlas matching using localized mutual information. Med Phys 35(4):1407–1417

    Article  PubMed  Google Scholar 

  45. Krygier MC, LaBonte T, Martinez C, Norris C, Sharma K, Collins LN, Roberts SA (2021) Quantifying the unknown impact of segmentation uncertainty on image-based simulations. Nat Commun 12(1):1–11

    Article  Google Scholar 

  46. Brüning J, Hellmeier F, Yevtushenko P, Kühne T, Goubergrits L (2018) Uncertainty quantification for non-invasive assessment of pressure drop across a coarctation of the aorta using CFD. Cardiovasc Eng Technol 9(4):582–596

    Article  PubMed  Google Scholar 

  47. Maher GD, Fleeter CM, Schiavazzi DE, Marsden AL (2021) Geometric uncertainty in patient-specific cardiovascular modeling with convolutional dropout networks. Comput Methods Appl Mech Eng 386

  48. Bryan R, Mohan PS, Hopkins A, Galloway F, Taylor M, Nair PB (2010) Statistical modelling of the whole human femur incorporating geometric and material properties. Med Eng Phys 32(1):57–65

    Article  PubMed  Google Scholar 

  49. Porziani S, Groth C, Waldman W, Biancolini ME (2021) Automatic shape optimisation of structural parts driven by BGM and RBF mesh morphing. Int J Mech Sci 189:105976

    Article  Google Scholar 

  50. Skala V (2017) RBF interpolation with CSRBF of large data sets. Procedia Comput Sci 108:2433–2437

    Article  Google Scholar 

  51. Hoeijmakers M, Huberts W, Rutten M, van de Vosse F (2021) The impact of shape uncertainty on aortic-valve pressure-drop computations. Int J Numer Methods Biomed Eng 37(10):3518

    Article  Google Scholar 

  52. Karmonik C, Diaz O, Klucznik R, Grossman RG, Zhang YJ, Britz G, Lv N, Huang Q (2015) Quantitative comparison of hemodynamic parameters from steady and transient CFD simulations in cerebral aneurysms with focus on the aneurysm ostium. J Neurointerv Surg 7(5):367–372

    Article  CAS  PubMed  Google Scholar 

  53. Condemi F, Campisi S, Viallon M, Troalen T, Xuexin G, Barker A, Markl M, Croisille P, Trabelsi O, Cavinato C et al (2017) Fluid-and biomechanical analysis of ascending thoracic aorta aneurysm with concomitant aortic insufficiency. Ann Biomed Eng 45(12):2921–2932

    Article  CAS  PubMed  Google Scholar 

  54. Gallo D, Gülan U, Di Stefano A, Ponzini R, Lüthi B, Holzner M, Morbiducci U (2014) Analysis of thoracic aorta hemodynamics using 3D particle tracking velocimetry and computational fluid dynamics. J Biomech 47(12):3149–3155

    Article  PubMed  Google Scholar 

  55. Reymond P, Crosetto P, Deparis S, Quarteroni A, Stergiopulos N (2013) Physiological simulation of blood flow in the aorta: comparison of hemodynamic indices as predicted by 3-D FSI, 3-D rigid wall and 1-D models. Med Eng Phys 35(6):784–791

    Article  PubMed  Google Scholar 

  56. Pons R, Guala A, Rodríguez-Palomares JF, Cajas J, Dux-Santoy L, Teixidó-Tura G, Molins JJ, Vázquez M, Evangelista A, Martorell J (2020) Fluid-structure interaction simulations outperform computational fluid dynamics in the description of thoracic aorta haemodynamics and in the differentiation of progressive dilation in Marfan syndrome patients. Roy Soc Open Sci 7(2):191752

    Article  CAS  Google Scholar 

  57. Suito H, Takizawa K, Huynh VQ, Sze D, Ueda T (2014) FSI analysis of the blood flow and geometrical characteristics in the thoracic aorta. Comput Mech 54(4):1035–1045

    Article  Google Scholar 

  58. Mendez V, Di Giuseppe M, Pasta S (2018) Comparison of hemodynamic and structural indices of ascending thoracic aortic aneurysm as predicted by 2-way FSI, CFD rigid wall simulation and patient-specific displacement-based FEA. Comput Biol Med 100:221–229

    Article  PubMed  Google Scholar 

Download references


We would like to thank Siemens Healthineers Company for providing the 4D flow MRI sequence. 


The authors did not receive support from any organization for the submitted work.

Author information

Authors and Affiliations



DMM-C: Methodology, Software, Conceptualization, Data Curation, Writing—Original Draft, Visualization, Investigation, Formal analysis. LG: Methodology, Software, Conceptualization, Writing—Original Draft, Visualization. AB: Supervision, Data Curation, Writing—Review and Editing. SL: Resources, Investigation. M-CM: Resources Alexandre Cochet: Writing—Review and Editing, Methodology. MR: Funding acquisition, Writing—Review and Editing, Methodology. SL: Conceptualization, Writing—Review and Editing. KA: Review and Editing. NJ: Review and Editing. LSA: Data Curation, Formal analysis. AL: Conceptualization, Writing—Review and Editing, Resources. OB: Supervision, Project administration. BP: Supervision, Conceptualization, Writing—Review and Editing, Validation.

Corresponding author

Correspondence to Benoit Presles.

Ethics declarations

Conflict of interest

The authors declare they have no conflict of interest. Financial interest: The authors have no relevant financial or non-financial interests to disclose.

Ethical standards

We declare that this research did not received any specific grant from funding agencies in the public, commercial or not-for-profit sectors.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Marin-Castrillon, D.M., Geronzi, L., Boucher, A. et al. Segmentation of the aorta in systolic phase from 4D flow MRI: multi-atlas vs. deep learning. Magn Reson Mater Phy 36, 687–700 (2023).

Download citation

  • Received:

  • Revised:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: