While the goal of fully-automatic ultrasound image segmentation remains elusive, we have shown how operator assistance can be exploited to produce fast, reliable and verifyable semi-automatic segmentation. Key features of our approach include high resolution segmentation from target points constrained by a B-spline snake, local statistical boundary models and on-the-fly training of the boundary models. 3D data sets can be segmented in a fraction of the time it would take to manually trace the boundaries in each frame. Further work could look into exploiting prior knowledge of an organ's shape [7] to further improve resilience to noise and reduce the amount of operator intervention. It would also be interesting to investigate other boundary indicators besides gradients and low order texture statistics.