An integrated approach to surface modeling in freehand three-dimensional echocardiography

We describe an integrated Bayesian solution to find a left ventricle model, including both epicardium and endocardium surfaces, from freehand 3-D echocardiographic images. The observed images and prior shape knowledge are combined to make the most consistent inference about unknown surface models using the maximum a posteriori rule. Typical model-based computer vision techniques divide the overall problem into two separate low and high-level subproblems. Unlike previous approaches, our approach unifies these two levels through a pixel class prediction mechanism. A putative surface model is generated from a catalog of 86 representative surface models. For each observed pixel, its appearance probability profile from different classes is first computed. Then the class predication probability profile is also computed, based only on the putative surface model. An optimal surface model has the best overall match between these two profiles for all the pixels. The probability models are obtained off-line by the expectation maximization algorithm from 20 training studies. Quantitative experimental results on 25 test studies show the advantage of the integrated approach.