Skip to main content

Prediction of pulmonary pressure after Glenn shunts by computed tomography–based machine learning models



This study aimed to develop non-invasive machine learning classifiers for predicting post–Glenn shunt patients with low and high risks of a mean pulmonary arterial pressure (mPAP) > 15 mmHg based on preoperative cardiac computed tomography (CT).


This retrospective study included 96 patients with functional single ventricle who underwent a bidirectional Glenn procedure between November 1, 2009, and July, 31, 2017. All patients underwent post-procedure CT, followed by cardiac catheterization. Overall, 23 morphologic parameters were manually extracted from cardiac CT images for each patient. The Mann-Whitney U or chi-square test was applied to select the most significant predictors. Six machine learning algorithms including logistic regression, Naive Bayes, random forest (RF), linear discriminant analysis, support vector machine, and K-nearest neighbor were used for modeling. These algorithms were independently trained on 100 train-validation random splits with a 3:1 ratio. Their average performance was evaluated by area under the curve (AUC), accuracy, sensitivity, and specificity.


Seven CT morphologic parameters were selected for modeling. RF obtained the best performance, with mean AUC of 0.840 (confidence interval [CI] 0.832–0.850) and 0.787 (95% CI 0.780–0.794); sensitivity of 0.815 (95% CI 0.797–0.833) and 0.778 (95% CI 0.767–0.788), specificity of 0.766 (95% CI 0.748–0.785) and 0.746 (95% CI 0.735–0.757); and accuracy of 0.782 (95% CI 0.771–0.793) and 0.756 (95% CI 0.748–0.764) in the training and validation cohorts, respectively.


The CT-based RF model demonstrates a good performance in the prediction of mPAP, which may reduce the need for right heart catheterization in post–Glenn shunt patients with suspected mPAP > 15 mmHg.

Key Points

• Twenty-three candidate descriptors were manually extracted from cardiac computed tomography images, and seven of them were selected for subsequent modeling.

• The random forest model presents the best predictive performance for pulmonary pressure among all methods.

• The computed tomography–based machine learning model could predict post–Glenn shunt pulmonary pressure non-invasively.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2



Anastomosis area


Inferior vena cava area


Left pulmonary artery area


Area of pulmonary artery below anastomosis


Right pulmonary artery area


Superior vena cava area


Area under the curve


Confidence interval


Computed tomography


Diameter of anastomosis major axis


Diameter of anastomosis minor axis


Diameter of inferior vena cava major axis


Diameter of inferior vena cava minor axis


Diameter of pulmonary artery major axis below anastomosis


Diameter of superior vena cava major axis


Diameter of superior vena cava minor axis


Major aortopulmonary collateral arteries


Mean pulmonary arterial pressure


Anastomosis perimeter


Inferior vena cava perimeter


Superior vena cava perimeter


Random forest


  1. 1.

    Tanoue Y, Kado H, Boku N et al (2007) Three hundred and thirty-three experiences with the bidirectional Glenn procedure in a single institute. Interact Cardiovasc Thorac Surg 6:97–101

    Article  Google Scholar 

  2. 2.

    Hosein RB, Clarke AJ, McGuirk SP et al (2007) Factors influencing early and late outcome following the Fontan procedure in the current era. The ‘Two Commandments?’. Eur J Cardiothorac Surg 31:344–352

    Article  Google Scholar 

  3. 3.

    Stern HJ (2010) Fontan “Ten Commandments” revisited and revised. Pediatr Cardiol 31:1131–1134

    Article  Google Scholar 

  4. 4.

    Mandel J, Poch D (2013) In the clinic. Pulmonary hypertension. Ann Intern Med 158:C1–C5

    Article  Google Scholar 

  5. 5.

    Hoeper MM, Lee SH, Voswinckel R et al (2006) Complications of right heart catheterization procedures in patients with pulmonary hypertension in experienced centers. J Am Coll Cardiol 48:2546–2552

    Article  Google Scholar 

  6. 6.

    Stern KW, McElhinney DB, Gauvreau K, Geva T, Brown DW (2011) Echocardiographic evaluation before bidirectional Glenn operation in functional single-ventricle heart disease: comparison to catheter angiography. Circ Cardiovasc Imaging 4:498–505

    Article  Google Scholar 

  7. 7.

    Shen Y, Wan C, Tian P et al (2014) CT-base pulmonary artery measurement in the detection of pulmonary hypertension: a meta-analysis and systematic review. Medicine (Baltimore) 93:e256

    Article  Google Scholar 

  8. 8.

    Janda S, Shahidi N, Gin K, Swiston J (2011) Diagnostic accuracy of echocardiography for pulmonary hypertension: a systematic review and meta-analysis. Heart 97:612–622

    Article  Google Scholar 

  9. 9.

    Wang N, Hu X, Liu C et al (2014) A systematic review of the diagnostic accuracy of cardiovascular magnetic resonance for pulmonary hypertension. Can J Cardiol 30:455–463

    CAS  Article  Google Scholar 

  10. 10.

    Chen X, Liu K, Wang Z et al (2015) Computed tomography measurement of pulmonary artery for diagnosis of COPD and its comorbidity pulmonary hypertension. Int J Chron Obstruct Pulmon Dis 10:2525–2533

    Article  Google Scholar 

  11. 11.

    Rajaram S, Swift AJ, Condliffe R et al (2015) CT features of pulmonary arterial hypertension and its major subtypes: a systematic CT evaluation of 292 patients from the ASPIRE Registry. Thorax 70:382–387

    CAS  Article  Google Scholar 

  12. 12.

    Grosse C, Grosse A (2010) CT findings in diseases associated with pulmonary hypertension: a current review. Radiographics 30:1753–1777

    Article  Google Scholar 

  13. 13.

    Abel E, Jankowski A, Pison C, Luc Bosson J, Bouvaist H, Ferretti GR (2012) Pulmonary artery and right ventricle assessment in pulmonary hypertension: correlation between functional parameters of ECG-gated CT and right-side heart catheterization. Acta Radiol 53:720–727

    Article  Google Scholar 

  14. 14.

    Siegel Y, Mirpuri T (2014) Pulmonary hypertension detection using dynamic and static measurable parameters on CT angiography. J Comput Assist Tomogr 38:586–590

    Article  Google Scholar 

  15. 15.

    Mahammedi A, Oshmyansky A, Hassoun PM, Thiemann DR, Siegelman SS (2013) Pulmonary artery measurements in pulmonary hypertension: the role of computed tomography. J Thorac Imaging 28:96–103

    Article  Google Scholar 

  16. 16.

    Goldstein BA, Navar AM, Carter RE (2017) Moving beyond regression techniques in cardiovascular risk prediction: applying machine learning to address analytic challenges. Eur Heart J 38:1805–1814

    PubMed  Google Scholar 

  17. 17.

    Krittanawong C, Zhang H, Wang Z, Aydar M, Kitai T (2017) Artificial intelligence in precision cardiovascular medicine. J Am Coll Cardiol 69:2657–2664

    Article  Google Scholar 

  18. 18.

    Henglin M, Stein G, Hushcha PV, Snoek J, Wiltschko AB, Cheng S (2017) Machine learning approaches in cardiovascular imaging. Circ Cardiovasc Imaging 10:e005614.

    Article  PubMed  PubMed Central  Google Scholar 

  19. 19.

    Darcy AM, Louie AK, Roberts LW (2016) Machine learning and the profession of medicine. JAMA 315:551–552

    CAS  Article  Google Scholar 

  20. 20.

    Hae H, Kang SJ, Kim WJ et al (2018) Machine learning assessment of myocardial ischemia using angiography: development and retrospective validation. PLoS Med 15:e1002693

    Article  Google Scholar 

  21. 21.

    Liu H, Juan YH, Wang Q et al (2014) Evaluation of malposition of the branch pulmonary arteries using cardiovascular computed tomography angiography. Eur Radiol 24:3300–3307

    Article  Google Scholar 

  22. 22.

    Yagi M, Taniguchi H, Kondoh Y et al (2017) CT-determined pulmonary artery to aorta ratio as a predictor of elevated pulmonary artery pressure and survival in idiopathic pulmonary fibrosis. Respirology 22:1393–1399

    Article  Google Scholar 

  23. 23.

    Nakata S, Imai Y, Takanashi Y et al (1984) A new method for the quantitative standardization of cross-sectional areas of the pulmonary arteries in congenital heart diseases with decreased pulmonary blood flow. J Thorac Cardiovasc Surg 88:610–619

    CAS  Article  Google Scholar 

  24. 24.

    Fiore AC, Tobin C, Jureidini S, Rahimi M, Kim ES, Schowengerdt K (2011) A comparison of the modified Blalock-Taussig shunt with the right ventricle-to-pulmonary artery conduit. Ann Thorac Surg 91:1479–1484 1484–1485

    Article  Google Scholar 

  25. 25.

    Jia Q, Cen J, Zhuang J et al (2017) Significant survival advantage of high pulmonary vein index and the presence of native pulmonary artery in pulmonary atresia with ventricular septal defect and major aortopulmonary collateral arteries: results from preoperative computed tomography angiography. Eur J Cardiothorac Surg 52:225–232

    Article  Google Scholar 

  26. 26.

    Maeda E, Akahane M, Kato N et al (2006) Assessment of major aortopulmonary collateral arteries with multidetector-row computed tomography. Radiat Med 24:378–383

    Article  Google Scholar 

  27. 27.

    Deo RC (2015) Machine Learning in Medicine. Circulation 132:1920–1930

    Article  Google Scholar 

  28. 28.

    Zhang Y, Zhang B, Liang F et al (2019) Radiomics features on non-contrast-enhanced CT scan can precisely classify AVM-related hematomas from other spontaneous intraparenchymal hematoma types. Eur Radiol 29:2157–2165

    Article  Google Scholar 

  29. 29.

    Ciritsis A, Rossi C, Eberhard M, Marcon M, Becker AS, Boss A (2019) Automatic classification of ultrasound breast lesions using a deep convolutional neural network mimicking human decision-making. Eur Radiol.

    Article  Google Scholar 

  30. 30.

    Lungu A, Swift AJ, Capener D, Kiely D, Hose R, Wild JM (2016) Diagnosis of pulmonary hypertension from magnetic resonance imaging-based computational models and decision tree analysis. Pulm Circ 6:181–190

    Article  Google Scholar 

  31. 31.

    Dawes T, de Marvao A, Shi W et al (2017) Machine learning of three-dimensional right ventricular motion enables outcome prediction in pulmonary hypertension: a cardiac MR imaging study. Radiology 283:381–390

    Article  Google Scholar 

  32. 32.

    Dennis A, Michaels AD, Arand P, Ventura D (2010) Noninvasive diagnosis of pulmonary hypertension using heart sound analysis. Comput Biol Med 40:758–764

    Article  Google Scholar 

  33. 33.

    Elgendi M, Bobhate P, Jain S et al (2018) The voice of the heart: vowel-like sound in pulmonary artery hypertension. Diseases 6:E26.

    Article  PubMed  Google Scholar 

  34. 34.

    Diller GP, Kempny A, Babu-Narayan SV, et al (2019) Machine learning algorithms estimating prognosis and guiding therapy in adult congenital heart disease: data from a single tertiary centre including 10 019 patients. Eur Heart J 40:1069–1077

    Article  Google Scholar 

  35. 35.

    Zhang B, He X, Ouyang F et al (2017) Radiomic machine-learning classifiers for prognostic biomarkers of advanced nasopharyngeal carcinoma. Cancer Lett 403:21–27

    CAS  Article  Google Scholar 

  36. 36.

    Nagasawa S, Al-Naamani E, Saeki A (2018) Computer-aided screening of conjugated polymers for organic solar cell: classification by random forest. J Phys Chem Lett 9:2639–2646

    CAS  Article  Google Scholar 

  37. 37.

    Yao Y, Li X, Liao B et al (2017) Predicting influenza antigenicity from Hemagglutintin sequence data based on a joint random forest method. Sci Rep 7:1545

    Article  Google Scholar 

  38. 38.

    Eshaghi A, Wottschel V, Cortese R et al (2016) Gray matter MRI differentiates neuromyelitis optica from multiple sclerosis using random forest. Neurology 87:2463–2470

    Article  Google Scholar 

  39. 39.

    Gray KR, Aljabar P, Heckemann RA, Hammers A, Rueckert D (2013) Random forest-based similarity measures for multi-modal classification of Alzheimer’s disease. Neuroimage 65:167–175

    Article  Google Scholar 

  40. 40.

    Esmaily-Moghadam M, Hsia TY, Marsden AL (2015) The assisted bidirectional Glenn: a novel surgical approach for first-stage single-ventricle heart palliation. J Thorac Cardiovasc Surg 149:699–705

    Article  Google Scholar 

Download references


This study was supported by the Key Program of Union of National Natural Science Foundation of China-Guangdong Province (U1401255), the Natural Science Foundation of Guangdong Province (2018A030313785), the Science and Technology Planning Project of Guangdong Province (2019B020230003, 2018B090944002, 2017A070701013, 2017B090904034, and 2017B030314109), the National key Research and Development Program (2018YFC1002600), and Guangdong peak project (DFJH201802).

Author information



Corresponding authors

Correspondence to Jijin Lin or Yuhao Dong.

Ethics declarations


The scientific guarantor of this publication is Yuhao Dong.

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

The author Dewen Zeng did statistical analyses.

Informed consent

Written informed consent was waived from all subjects (patients) in this study.

Ethical approval

Institutional Review Board approval was obtained.


• Retrospective

• Diagnostic study or prognostic

• Performed at one institution

Additional information

Publisher’s note

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

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Huang, L., Li, J., Huang, M. et al. Prediction of pulmonary pressure after Glenn shunts by computed tomography–based machine learning models. Eur Radiol 30, 1369–1377 (2020).

Download citation


  • Heart diseases
  • Lung
  • Pressure
  • Multi-detector computed tomography
  • Machine learning