A Fully Automatic Mid-Diastole Frame Detection in 2D Echocardiography Sequences

Mitral valve stenosis (MS) is one of the most prevalent heart valve diseases. The mitral valve orifice area (MVA) is a reliable measure for evaluating the MS severity. This measure is typically obtained by applying planimetry methods to the echocardiography video frames in a parasternal short axis view (PSAX). From the diagnostic perspective, the most proper frame for measuring the MVA is the mid-diastole frame. Since manual methods for localizing this frame within a sequence of recorded echocardiography frames are time-consuming and user-dependent, a novel three-stage approach is proposed. Firstly, the mitral valve orifice (MVO) region is detected automatically for each frame using the Circular Hough transform (CHT) leading to a lower computational cost and an increased accuracy. Secondly, for each of the detected MVO regions, a data dimensionality reduction (DR) method is applied to map it into a single point in a two-dimensional (2D) space. Finally, a curve is generated based on Euclidean distances between consecutive points in the 2D space. The curve analysis allows for the detection of the location of the mid-diastole frame. The proposed algorithm was validated using echocardiographic video sequences collected from 11 different participants including 7 individuals diagnosed with mitral valve stenosis and 4 control non-stenosis individuals. The mid-diastole frame location estimated by an expert cardiologist was used as the gold standard for assessing the evaluation results. The performance of different DR methods including the linear principal component analysis (LPCA) and the kernel PCA (with a polynomial or a Gaussian kernel) was evaluated. The PCA with Gaussian kernel produced the best results with the average difference between the proposed method and the gold standard equal to 0.57 frames in MS and 0.50 frames in non-MS cases.

[1]  Josef Kittler,et al.  A Comparative Study of Hough Transform Methods for Circle Finding , 1989, Alvey Vision Conference.

[2]  Neil Davey,et al.  Analysis of linear and nonlinear dimensionality reduction methods for gender classification of face images , 2005, Int. J. Syst. Sci..

[3]  N. Kachenoura,et al.  Automatic Detection of End Systole within a Sequence of Left Ventricular Echocardiographic Images using Autocorrelation and Mitral Valve Motion Detection , 2007, 2007 29th Annual International Conference of the IEEE Engineering in Medicine and Biology Society.

[4]  Eric O. Postma,et al.  Dimensionality Reduction: A Comparative Review , 2008 .

[5]  J. Chambers,et al.  Echocardiographic assessment of valve stenosis: EAE/ASE recommendations for clinical practice. , 2009, European journal of echocardiography : the journal of the Working Group on Echocardiography of the European Society of Cardiology.

[6]  F. Flachskampf,et al.  Recommendations for the evaluation of left ventricular diastolic function by echocardiography. , 2008, European journal of echocardiography : the journal of the Working Group on Echocardiography of the European Society of Cardiology.

[7]  H. Behnam,et al.  Automatic detection of end-diastole and end-systole from echocardiography images using manifold learning. , 2010, Physiological measurement.

[8]  H. Behnam,et al.  Automatic detection of end systole and end diastole within a sequence of 2-D echocardiographic images using modified Isomap algorithm , 2011, 2011 1st Middle East Conference on Biomedical Engineering.

[9]  A. Omran,et al.  Echocardiography in mitral stenosis. , 2011, Journal of the Saudi Heart Association.

[10]  M. Salarifar,et al.  Mitral annular calcification predicts immediate results of percutaneous transvenous mitral commissurotomy , 2011, Cardiovascular ultrasound.

[11]  Quan Wang,et al.  Kernel Principal Component Analysis and its Applications in Face Recognition and Active Shape Models , 2012, ArXiv.

[12]  S. Min,et al.  Discrepancy between mitral valve areas measured by two-dimensional planimetry and three-dimensional transoesophageal echocardiography in patients with mitral stenosis , 2012, Heart.

[13]  T. Gillebert,et al.  Echo-Doppler assessment of diastole: flow, function and haemodynamics , 2012, Heart.

[14]  Luminita Moraru,et al.  Cardiac Cycle Phase Estimation in 2-D Echocardiographic Images Using an Artificial Neural Network , 2013, IEEE Transactions on Biomedical Engineering.

[15]  Michel Verleysen,et al.  Nonlinear Dimensionality Reduction for Visualization , 2013, ICONIP.

[16]  A. Saxena.  Echocardiographic diagnosis of chronic rheumatic valvular lesions. , 2013, Global heart.

[17]  Steven Lawrence Fernandes,et al.  Performance Analysis of PCA-based and LDA- based Algorithms for Face Recognition , 2013, SiPS 2013.

[18]  R. Siegel,et al.  Management of mitral stenosis using 2D and 3D echo-Doppler imaging. , 2013, JACC. Cardiovascular imaging.

[19]  S. Sundaramurthy,et al.  Cardiac Cycle Phase Detection in Echocardiography Images Using ANN , 2014, 2014 International Conference on Intelligent Computing Applications.

[20]  Yambem Jina Chanu,et al.  Image Segmentation Using K -means Clustering Algorithm and Subtractive Clustering Algorithm , 2015 .

[21]  J. Hung,et al.  Standard transthoracic echocardiography and transesophageal echocardiography views of mitral pathology that every surgeon should know. , 2015, Annals of cardiothoracic surgery.

[22]  Deshmukh Sachin,et al.  Dimensionality Reduction and Classification through PCA and LDA , 2015 .

[23]  S Chetan.  Circular Hough Transform For Detecting And Measuring Circles of Object , 2015 .

[24]  K. K. Sahu,et al.  Image Compression Methods using Dimension Reduction and Classification through PCA and LDA : A Review , 2016 .

[25]  B. Surendiran,et al.  Dimensionality reduction using Principal Component Analysis for network intrusion detection , 2016 .

[26]  Fan Yang,et al.  Convolutional Neural Network for the Detection of End-Diastole and End-Systole Frames in Free-Breathing Cardiac Magnetic Resonance Imaging , 2017, Comput. Math. Methods Medicine.

[27]  Lopamudra Dey,et al.  Outlier Detection and Removal Algorithm in K-Means and Hierarchical Clustering , 2017 .

[28]  S. A. Khonsary.  Guyton and Hall: Textbook of Medical Physiology , 2017, Surgical Neurology International.

[29]  D. Francis,et al.  Automatic detection of end‐diastolic and end‐systolic frames in 2D echocardiography , 2017, Echocardiography.

[30]  Hamidullah Binol,et al.  Ensemble Learning Based Multiple Kernel Principal Component Analysis for Dimensionality Reduction and Classification of Hyperspectral Imagery , 2018, Mathematical Problems in Engineering.

[31]  M. Yacoub,et al.  A novel method to measure mitral valve area in patients with rheumatic mitral stenosis using three‐dimensional transesophageal echocardiography: Feasibility and validation , 2018, Echocardiography.

[32]  R. Siegel,et al.  Imaging for Mitral Interventions: Methods and Efficacy. , 2018, JACC. Cardiovascular imaging.