An approach for contour detection of human kidneys from ultrasound images using Markov random fields and active contours Export

@article{citeulike:409964, abstract = {In this paper, a novel method for the boundary detection of human kidneys from three dimensional (3D) ultrasound (US) is proposed. The inherent difficulty of interpretation of such images, even by a trained expert, makes the problem unsuitable for classical methods. The method here proposed finds the kidney contours in each slice. It is a probabilistic Bayesian method. The prior defines a Markov field of deformations and imposes the restriction of contour smoothness. The likelihood function imposes a probabilistic behavior to the data, conditioned to the contour position. This second function, which is also Markov, uses an empirical model of distribution of the echographical data and a function of the gradient of the data. The model finally includes, as a volumetric extension of the prior, a term that forces smoothness along the depth coordinate. The experiments that have been carried out on echographies from real patients validate the model here proposed. A sensitivity analysis of the model parameters has also been carried out.}, author = {Martin-Fernandez, Marcos and Alberola-Lopez, Carlos}, citeulike-article-id = {409964}, citeulike-linkout-0 = {http://dx.doi.org/10.1016/j.media.2004.05.001}, citeulike-linkout-1 = {http://www.sciencedirect.com/science/article/B6W6Y-4CSG1S8-1/2/9b94b0c2260d62b95057f6eaaa7e40bc}, doi = {10.1016/j.media.2004.05.001}, journal = {Medical Image Analysis}, keywords = {contour, image, medical, mrf}, month = {February}, number = {1}, pages = {1--23}, posted-at = {2005-11-28 10:32:04}, priority = {2}, title = {An approach for contour detection of human kidneys from ultrasound images using Markov random fields and active contours}, url = {http://dx.doi.org/10.1016/j.media.2004.05.001}, volume = {9}, year = {2005} }

TY - JOUR ID - citeulike:409964 L3 - citeulike-article-id:409964 N2 - In this paper, a novel method for the boundary detection of human kidneys from three dimensional (3D) ultrasound (US) is proposed. The inherent difficulty of interpretation of such images, even by a trained expert, makes the problem unsuitable for classical methods. The method here proposed finds the kidney contours in each slice. It is a probabilistic Bayesian method. The prior defines a Markov field of deformations and imposes the restriction of contour smoothness. The likelihood function imposes a probabilistic behavior to the data, conditioned to the contour position. This second function, which is also Markov, uses an empirical model of distribution of the echographical data and a function of the gradient of the data. The model finally includes, as a volumetric extension of the prior, a term that forces smoothness along the depth coordinate. The experiments that have been carried out on echographies from real patients validate the model here proposed. A sensitivity analysis of the model parameters has also been carried out. IS - 1 JF - Medical Image Analysis EP - 23 TI - An approach for contour detection of human kidneys from ultrasound images using Markov random fields and active contours VL - 9 SP - 1 KW - contour KW - image KW - medical KW - mrf AU - Martin-Fernandez, Marcos AU - Alberola-Lopez, Carlos PY - 2005/02// UR - http://dx.doi.org/10.1016/j.media.2004.05.001 ER -