Automatic Detection of Left Ventricular Ejection Fraction Based on Fully Convolutional Networks

doi:10.12068/j.issn.1005-3026.2018.11.011

Abstract

Abstract: An automatic estimation method of left ventricular ejection fraction based on fully convolutional networks(FCN)is proposed. The left ventricle in magnetic resonance images(MRI) of heart is segmented using FCN. Furthermore， the volume of left ventricle can be calculated at each phase in a heart beat cycle. Finally， the volume of end-systolic and end-diastolic are extracted respectively to deduce the left ventricular ejection fraction. 700 sets of images are used for training the networks and 400 sets for testing. The final results agree well with the ejection fraction(EF)gold standard provided by the American National Institutes of Health and Children′s National Medical Center. The accuracy of the proposed method achieves 89.8%， which is within an acceptable range.

Key words: fully convolutional networks(FCN), ejection fraction(EF), magnetic resonance images(MRI), left ventricular segmentation

CLC Number:

TP391.41

XU Li-sheng， ZHANG Shu-qi， NIU Xiao， XU Yang. Automatic Detection of Left Ventricular Ejection Fraction Based on Fully Convolutional Networks[J]. Journal of Northeastern University Natural Science, 2018, 39(11): 1572-1576.

References

[1]陈伟伟，高润霖，刘力生，等.中国心血管病报告2016［J］.中国循环杂志，2017，32(6):521-530.(Chen Wei-wei，Gao Run-lin，Liu Li-sheng，et al.Cardiovascular diseases in China 2016［J］.Chinese Journal of Recurrence，2017，32(6):521-530.)
[2]Buckert D，Kelle S，Buss S，et al.Left ventricular ejection fraction and presence of myocardial necrosis assessed by cardiac magnetic resonance imaging correctly risk stratify patients with stable coronary artery disease:a multi-center all-comers trial［J］.Clinical Research in Cardiology，2016，106(3):1-11.
[3]Pedrosa J，Heyde B，Heeren L，et al.Automatic short axis orientation of the left ventricle in 3D ultrasound recordings［C］//SPIE Medical Imaging.San Diego，2016:97900E.
[4]Khan A，Iskandar D N F A，Ujir H，et al.Automatic segmentation of CMRIs for LV contour detection［C］//9th International Conference on Robotic，Vision，Signal Processing and Power Applications.Singapore，2017:313-321.
[5]Guo Z Z，Tan W，Xu L，et al.A level set segmentation method of the right ventricle in cardiac MRI images based on motion intensity［C］//Proceedings of the 17th China-Japan-Korea Joint Symposium on Medical Informatics.Seoul，2016:62-65.
[6]Luo Y，Yang B，Xu L，et al.Segmentation of the left ventricle in cardiac MRI using an ELM model［C］//Proceedings of ELM-2015.Berlin，2016:147-159.
[7]Luo Y，Yang B，Xu L，et al.Segmentation of the left ventricle in cardiac MRI using a hierarchical extreme learning machine model［J］.International Journal of Machine Learning & Cybernetics，2017，9:1-11.
[8]Cocosco C，Niessen W T，Vonken E，et al.Automatic image-driven segmentation of the ventricles in cardiac cine MRI［J］. Journal of Magnetic Resonance Imaging，2008，28(2):366-374.
[9]Billet F，Sermesant M，Delingette H，et al.Cardiac motion recovery and boundary conditions estimation by coupling an electromechanical model and cine-MRI data［C］//International Conference of on Functional Imaging and Modeling of the Heart.Nice，2009:376-385.
[10]Zhang H，Wahle A，Johnson R K，et al.4-D Cardiac MR image analysis:left and right ventricular morphology and function［J］.IEEE Transactions on Medical Imaging，2010，29(2):350-364.
[11]Xiong C，Zhao X，Tang D，et al.Conditional convolutional neural network for modality-aware face recognition［C］//IEEE International Conference on Computer Vision.New York，2016:3667-3675.
[12]Shelhamer E，Long J，Darrell T，et al.Fully convolutional networks for semantic segmentation［J］.IEEE Transactions on Pattern Analysis & Machine Intelligence，2017，39(4):640-651.
[13]Petitjean C，Dacher J N.A review of segmentation methods in short axis cardiac MR images［J］.Medical Image Analysis，2011，15(2):169-184.
[14]La G A，Claessen G，Vand B A，et al.Cardiac MRI:a new gold standard for ventricular volume quantification during high-intensity exercise［J］.Circulation Cardiovascular Imaging，2013，6(2):329-338.
[15]Avendi M R，Kheradvar A，Jafarkhani H，et al.A combined deep-learning and deformable-model approach to fully automatic segmentation of the left ventricle in cardiac MRI［J］.Medical Image Analysis，2016，30:108-119.