Fully automatic left ventricular boundary extraction in echocardiographic images
暂无分享,去创建一个
Describes a fully automatic, radial search based LV boundary extraction algorithm for echocardiographic images. Neural network classifiers are used with new input feature vectors to detect the LV centre and LV edge points. The centre detection stage combines these neural classifiers with knowledge based techniques to refine the centre estimate. Knowledge guided snakes are developed to extract the epicardial and endocardial boundaries by linking candidate edge points. The snakes' energy functions are minimised using a new two stage dynamic programming method, which is several times faster than the existing method. Knowledge is used to guide the snakes through the edge points improving their accuracy and robustness.
[1] John J. Soraghan,et al. Detection of echocardiographic left ventricle boundaries using neural networks , 1993, Proceedings of Computers in Cardiology Conference.
[2] D. Adam,et al. Automatic ventricular cavity boundary detection from sequential ultrasound images using simulated annealing. , 1989, IEEE transactions on medical imaging.
[3] Francoise J. Preteux,et al. Deformable markers: mathematical morphology for active contour models control , 1991, Optics & Photonics.