Bayesian contour estimation of the left ventricle in ultrasound images of the heart

The problem of the left ventricle contour estimation in ultrasound heart images is considered. A Bayesian approach to the problem is presented. The left ventricle contour is modelled as one-dimensional noncausal Gauss-Markov random field. The algorithm estimates the contour by computing the statistical model parameters of the inside and the outside of the left ventricle and by minimizing an energy function that describes the contour. The result is an algorithm based on MAP criteria. The method exhibits robustness against the poor quality of the ultrasound images