Structured Random Forests for Myocardium Delineation in 3D Echocardiography

Delineation of myocardium borders from 3D echocardiography is a critical step for the diagnosis of heart disease. Following the approach of myocardium segmentation as a contour finding task, recent work has shown effective methods to interpret endocardial edge information in the left ventricle. Nevertheless, these methods are still prone to preserve irrelevant edge responses and would struggle to overcome chief ventricle anatomical challenges. In this paper we adapt Structured Random Forests, borrowed from computer vision, for fast and robust myocardium edge detection. This method is evaluated on a dataset composed of short-axis slices from 25 End-Diastolic echocardiography volumes. Results show that the proposed ensemble model outperforms standard intensity-based and local phase-based edge detectors, while removing or significantly suppressing irrelevant edges triggered by ultrasound image artefacts and blood pool anatomical structures.

[1]  Joseph J. Lim,et al.  Sketch Tokens: A Learned Mid-level Representation for Contour and Object Detection , 2013, 2013 IEEE Conference on Computer Vision and Pattern Recognition.

[2]  Xiaofeng Ren,et al.  Multi-scale Improves Boundary Detection in Natural Images , 2008, ECCV.

[3]  Sebastian Nowozin,et al.  Structured Prediction and Learning in Computer Vision , 2011 .

[4]  Antonio Criminisi,et al.  Decision Forests: A Unified Framework for Classification, Regression, Density Estimation, Manifold Learning and Semi-Supervised Learning , 2012, Found. Trends Comput. Graph. Vis..

[5]  Andrew J. Davison,et al.  Active Matching , 2008, ECCV.

[6]  Sebastian Nowozin,et al.  Structured Learning and Prediction in Computer Vision , 2011, Found. Trends Comput. Graph. Vis..

[7]  C. Lawrence Zitnick,et al.  Structured Forests for Fast Edge Detection , 2013, 2013 IEEE International Conference on Computer Vision.

[8]  John F. Canny,et al.  A Computational Approach to Edge Detection , 1986, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[9]  J. Alison Noble,et al.  Delineating anatomical boundaries using the boundary fragment model , 2013, Medical Image Anal..

[10]  Kashif Rajpoot,et al.  Local-phase based 3D boundary detection using monogenic signal and its application to real-time 3-D echocardiography images , 2009, 2009 IEEE International Symposium on Biomedical Imaging: From Nano to Macro.