Skip to main content
SpringerLink
Log in

Fully automated intracardiac 4D flow MRI post-processing using deep learning for biventricular segmentation

  • Imaging Informatics and Artificial Intelligence
  • Published:
European Radiology Aims and scope Submit manuscript

Abstract

Objectives

4D flow MRI allows for a comprehensive assessment of intracardiac blood flow, useful for assessing cardiovascular diseases, but post-processing requires time-consuming ventricular segmentation throughout the cardiac cycle and is prone to subjective errors. Here, we evaluate the use of automatic left and right ventricular (LV and RV) segmentation based on deep learning (DL) network that operates on short-axis cine bSSFP images.

Methods

A previously published DL network was fine-tuned via retraining on a local database of 106 subjects scanned at our institution. In 26 test subjects, the ventricles were segmented automatically by the network and manually by 3 human observers on bSSFP MRI. The bSSFP images were then registered to the corresponding 4D flow images to apply the segmentation to 4D flow velocity data. Dice coefficients and the relative deviation between measurements (automatic vs. manual and interobserver manual) of various hemodynamic parameters were assessed.

Results

The automated segmentation resulted in similar Dice scores (LV: 0.92, RV: 0.86) and lower relative deviations from manual segmentation in left ventricular (LV) average kinetic energy (KE) (8%) and RV KE (15%) than the Dice scores (LV: 0.91, RV: 0.87) and relative deviations between manual segmentation observers (LV KE: 11%, p = 0.01; RV KE: 19%, p = 0.03).

Conclusions

The automated post-processing method using deep learning resulted in hemodynamic measurements that differ from a manual observer’s measurements equally or less than the variation between manual observers. This approach can be used to decrease post-processing time on intraventricular 4D flow data and mitigate interobserver variability.

Key Points

• Our proposed method allows for fully automated post-processing of intraventricular 4D flow MRI data.

• Our method resulted in hemodynamic measurements that matched those derived from manual segmentation equally as well as interobserver variability.

• Our method can be used to greatly accelerate intraventricular 4D flow post-processing and improve interobserver repeatability.

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

Access this article

Log in via an institution

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

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4

Similar content being viewed by others

Abbreviations

4D flow MRI:

4-Dimensional velocity-encoded MRI

bSSFP:

Balanced steady-state free procession

CMR:

Cardiovascular magnetic resonance

DL:

Deep learning

EDV:

End diastolic volume

ESV:

End systolic volume

KE:

Kinetic energy

LV:

Left ventricle

RV:

Right ventricle

UKBB:

UK Biobank

References

  1. Dyverfeldt P, Bissell M, Barker AJ et al (2015) 4D flow cardiovascular magnetic resonance consensus statement. J Cardiovasc Magn Reson 17:72. https://doi.org/10.1186/s12968-015-0174-5

    Article  PubMed  PubMed Central  Google Scholar 

  2. Brandts A, Bertini M, Van Dijk EJ et al (2011) Left ventricular diastolic function assessment from three-dimensional three-directional velocity-encoded MRI with retrospective valve tracking. J Magn Reson Imaging 33:312–319. https://doi.org/10.1002/jmri.22424

    Article  PubMed  Google Scholar 

  3. Kamphuis VP, van der Palen RLF, de Koning PJH et al (2018) In-scan and scan–rescan assessment of LV in- and outflow volumes by 4D flow MRI versus 2D planimetry. J Magn Reson Imaging 47:511–522. https://doi.org/10.1002/jmri.25792

    Article  PubMed  Google Scholar 

  4. Hanneman K, Sivagnanam M, Nguyen ET et al (2014) Magnetic resonance assessment of pulmonary (QP) to systemic (QS) flows using 4D phase-contrast imaging: pilot study comparison with standard through-plane 2D phase-contrast imaging. Acad Radiol 21:1002–1008. https://doi.org/10.1016/j.acra.2014.04.012

    Article  PubMed  Google Scholar 

  5. Gabbour M, Schnell S, Jarvis K et al (2015) 4-D flow magnetic resonance imaging: blood flow quantification compared to 2-D phase-contrast magnetic resonance imaging and Doppler echocardiography. Pediatr Radiol 45:804–813. https://doi.org/10.1007/s00247-014-3246-z

    Article  PubMed  Google Scholar 

  6. Jarvis K, Schnell S, Barker AJ et al (2016) Evaluation of blood flow distribution asymmetry and vascular geometry in patients with Fontan circulation using 4-D flow MRI. Pediatr Radiol 46:1507–1519. https://doi.org/10.1007/s00247-016-3654-3

    Article  PubMed  PubMed Central  Google Scholar 

  7. Valverde I, Nordmeyer S, Uribe S et al (2012) Systemic-to-pulmonary collateral flow in patients with palliated univentricular heart physiology: measurement using cardiovascular magnetic resonance 4D velocity acquisition. J Cardiovasc Magn Reson 14:1–11. https://doi.org/10.1186/1532-429X-14-25

    Article  Google Scholar 

  8. Garcia J, Barker AJ, Markl M (2019) The role of imaging of flow patterns by 4D flow MRI in aortic stenosis. JACC Cardiovasc Imaging 12:252–266. https://doi.org/10.1016/j.jcmg.2018.10.034

    Article  PubMed  Google Scholar 

  9. Carlsson M, Heiberg E, Toger J, Arheden H (2012) Quantification of left and right ventricular kinetic energy using four-dimensional intracardiac magnetic resonance imaging flow measurements. AJP Hear Circ Physiol 302:H893–H900. https://doi.org/10.1152/ajpheart.00942.2011

    Article  CAS  Google Scholar 

  10. Eriksson J, Carlhäll C, Dyverfeldt P et al (2010) Semi-automatic quantification of 4D left ventricular blood flow. J Cardiovasc Magn Reson 12:9. https://doi.org/10.1186/1532-429X-12-9

    Article  PubMed  PubMed Central  Google Scholar 

  11. Fredriksson AG, Zajac J, Eriksson J et al (2011) 4-D blood flow in the human right ventricle. Am J Physiol Heart Circ Physiol 301:2344–2350. https://doi.org/10.1152/ajpheart.00622.2011

    Article  CAS  Google Scholar 

  12. Eriksson J, Dyverfeldt P, Engvall J et al (2011) Quantification of presystolic blood flow organization and energetics in the human left ventricle. AJP Hear Circ Physiol 300:H2135–H2141. https://doi.org/10.1152/ajpheart.00993.2010

    Article  CAS  Google Scholar 

  13. Rutkowski DR, Barton G, François CJ et al (2019) Analysis of cavopulmonary and cardiac flow characteristics in fontan patients: comparison with healthy volunteers. J Magn Reson Imaging 49:1786–1799. https://doi.org/10.1002/jmri.26583

    Article  PubMed  PubMed Central  Google Scholar 

  14. Corrado PA, Macdonald JA, François CJ et al (2019) Reduced regional flow in the left ventricle after anterior acute myocardial infarction: a case control study using 4D flow MRI. BMC Med Imaging 19:101. https://doi.org/10.1186/s12880-019-0404-7

    Article  PubMed  PubMed Central  Google Scholar 

  15. Hussaini SF, Rutkowski DR, Roldan-Alzate A, Francois CJ (2017) Left and right ventricular kinetic energy using time-resolved versus time-average ventricular volumes. J Magn Reson Imaging 45:821–828. https://doi.org/10.1002/jmri.25416

    Article  PubMed  Google Scholar 

  16. Fenster BE, Browning J, Schroeder JD et al (2015) Vorticity is a marker of right ventricular diastolic dysfunction. Am J Physiol Heart Circ Physiol 309:H1087–H1093. https://doi.org/10.1152/ajpheart.00278.2015

    Article  CAS  PubMed  Google Scholar 

  17. Eriksson J, Bolger AF, Ebbers T, Carlhäll CJ (2013) Four-dimensional blood flow-specific markers of LV dysfunction in dilated cardiomyopathy. Eur Heart J Cardiovasc Imaging 14:417–424. https://doi.org/10.1093/ehjci/jes159

    Article  PubMed  Google Scholar 

  18. Garg P, Crandon S, Swoboda PP et al (2018) (2018) Left ventricular blood flow kinetic energy after myocardial infarction - insights from 4D flow cardiovascular magnetic resonance. J Cardiovasc Magn Reson 201(20):61. https://doi.org/10.1186/s12968-018-0483-6

    Article  Google Scholar 

  19. Fredriksson AG, Svalbring E, Eriksson J et al (2016) 4D flow MRI can detect subtle right ventricular dysfunction in primary left ventricular disease. J Magn Reson Imaging 43:558–565. https://doi.org/10.1002/jmri.25015

    Article  PubMed  Google Scholar 

  20. Bock J, Frydrychowicz A, Stalder AF et al (2010) 4D phase contrast MRI at 3 T: Effect of standard and blood-pool contrast agents on SNR, PC-MRA, and blood flow visualization. Magn Reson Med 63:330–338. https://doi.org/10.1002/mrm.22199

    Article  PubMed  Google Scholar 

  21. Berhane H, Scott M, Elbaz M et al (2020) Fully automated 3D aortic segmentation of 4D flow MRI for hemodynamic analysis using deep learning. Magn Reson Med 1–15. https://doi.org/10.1002/mrm.28257

  22. Cibis M, Bustamante M, Eriksson J et al (2017) Creating hemodynamic atlases of cardiac 4D flow MRI. J Magn Reson Imaging 46:1389–1399. https://doi.org/10.1002/jmri.25691

    Article  PubMed  PubMed Central  Google Scholar 

  23. Bustamante M, Gupta V, Forsberg D et al (2018) Automated multi-atlas segmentation of cardiac 4D flow MRI. Med Image Anal 49:128–140. https://doi.org/10.1016/j.media.2018.08.003

    Article  PubMed  Google Scholar 

  24. Gupta V, Bustamante M, Fredriksson A et al (2018) Improving left ventricular segmentation in four-dimensional flow MRI using intramodality image registration for cardiac blood flow analysis. Magn Reson Med 79:554–560. https://doi.org/10.1002/mrm.26674

    Article  PubMed  Google Scholar 

  25. Bai W, Sinclair M, Tarroni G et al (2018) Automated cardiovascular magnetic resonance image analysis with fully convolutional networks. J Cardiovasc Magn Reson 20:1–12

    Article  Google Scholar 

  26. Tufvesson J, Hedstrom E, Steding-Ehrenborg K et al (2015) Validation and development of a new automatic algorithm for time-resolved segmentation of the left ventricle in magnetic resonance imaging. Biomed Res Int 2015:1–12. https://doi.org/10.1155/2015/970357

    Article  Google Scholar 

  27. Schulz-Menger J, Bluemke DA, Bremerich J et al (2013) Standardized image interpretation and post processing in cardiovascular magnetic resonance: Society for Cardiovascular Magnetic Resonance (SCMR) Board of Trustees Task Force on Standardized Post Processing. J Cardiovasc Magn Reson 15:1–19. https://doi.org/10.1186/1532-429X-15-35

    Article  Google Scholar 

  28. Schulz-Menger J, Bluemke DA, Bremerich J et al (2020) Standardized image interpretation and post-processing in cardiovascular magnetic resonance - 2020 update: Society for Cardiovascular Magnetic Resonance (SCMR): Board of Trustees Task Force on Standardized Post-Processing. J Cardiovasc Magn Reson 22:1–22. https://doi.org/10.1186/s12968-020-00610-6

    Article  Google Scholar 

  29. Gu T, Korosec FR, Block WF et al (2005) PC VIPR: a high-speed 3D phase-contrast method for flow quantification and high-resolution angiography. AJNR Am J Neuroradiol 26(743–749) https://doi.org/26/4/743 [pii]

  30. Avants BB, Tustison NJ, Stauffer M et al (2014) The insight ToolKit image registration framework. Front Neuroinform 8:1–13. https://doi.org/10.3389/fninf.2014.00044

    Article  Google Scholar 

  31. Shrout PE, Fleiss JL (1979) Intraclass correlations: uses in assessing rater reliability. Psychol Bull 86:420–428. https://doi.org/10.1037//0033-2909.86.2.420

    Article  CAS  PubMed  Google Scholar 

  32. Bland JM, Altman DG (2010) Statistical methods for assessing agreement between two methods of clinical measurement. Int J Nurs Stud 47:931–936. https://doi.org/10.1016/j.ijnurstu.2009.10.001

    Article  Google Scholar 

  33. Bernard O, Lalande A, Zotti C et al (2018) Deep learning techniques for automatic MRI cardiac multi-structures segmentation and diagnosis: is the problem solved? IEEE Trans Med Imaging 37:2514–2525. https://doi.org/10.1109/TMI.2018.2837502

    Article  PubMed  Google Scholar 

  34. Isensee F, Jaeger PF, Full PM, et al (2018) Automatic cardiac disease assessment on cine-MRI via time-series segmentation and domain specific features. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics) 10663 LNCS:120–129. https://doi.org/10.1007/978-3-319-75541-0_13

  35. Bartoli A, Fournel J, Bentatou Z et al (2021) Deep learning–based automated segmentation of left ventricular trabeculations and myocardium on cardiac MR images: a feasibility study. Radiol Artif Intell 3:e200021. https://doi.org/10.1148/ryai.2020200021

    Article  PubMed  Google Scholar 

  36. Albà X, Lekadir K, Pereañez M et al (2018) Automatic initialization and quality control of large-scale cardiac MRI segmentations. Med Image Anal 43:129–141. https://doi.org/10.1016/j.media.2017.10.001

    Article  PubMed  Google Scholar 

  37. Guo S, Xu L, Feng C et al (2021) Multi-level semantic adaptation for few-shot segmentation on cardiac image sequences. Med Image Anal 73:102170. https://doi.org/10.1016/j.media.2021.102170

    Article  PubMed  Google Scholar 

Download references

Funding

The authors state that this work has not received any funding.

Author information

Authors and Affiliations

  1. University of Wisconsin-Madison, 1111 Highland Ave, Madison, WI, 53705, USA

    Philip A. Corrado, Andrew L. Wentland, Jitka Starekova, Archana Dhyani & Oliver Wieben

  2. UT Southwestern Medical Center, 5323 Harry Hines Blvd., Dallas, TX, 75390, USA

    Kara N. Goss

Authors
  1. Philip A. Corrado
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Andrew L. Wentland
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Jitka Starekova
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Archana Dhyani
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Kara N. Goss
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Oliver Wieben
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Philip A. Corrado.

Ethics declarations

Guarantor

The scientific guarantor of this publication is Philip A. Corrado.

Conflict of interest

The authors of this manuscript declare relationships with the following companies: the University of Wisconsin receives research support from GE Healthcare and Bracco Diagnostics, outside the submitted work.

Statistics and biometry

No complex statistical methods were necessary for this paper.

Informed consent

Written informed consent was obtained from all subjects in this study.

Ethical approval

Institutional Review Board approval was obtained.

Study subjects or cohorts overlap

Some study subjects or cohorts have been previously reported in several other studies, mainly looking at the flow characteristics of the subject populations reported on. This methodology study is distinct from that work; however, the 4D flow MR images reported on herein have also been utilized in those studies, which are listed below:

Methodology

• Retrospective

• observational

• performed at one institution

Additional information

Publisher’s note

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

Supplementary information

ESM 1

(DOCX 475 kb)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Corrado, P.A., Wentland, A.L., Starekova, J. et al. Fully automated intracardiac 4D flow MRI post-processing using deep learning for biventricular segmentation. Eur Radiol 32, 5669–5678 (2022). https://doi.org/10.1007/s00330-022-08616-7

Download citation

  • Received:

  • Revised:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00330-022-08616-7

Keywords

Search

Navigation