3-D Active Shape Image Segmentation Using a Scale Model

In this paper, we propose an active shape image segmentation method for three-dimensional (3-D) medical images. The approach consists of a generation method of a 3-D shape model and a segmentation method using a scale model. The 3-D shape generation method uses a tetrahedron algorithm for landmarking based on the view of geometry. After generating the 3-D model, we extend the shape model training and gray-level model training of two-dimensional (2-D) active shape models (ASMs). We use the integrated modeling process with scale and gray-level models for the appearance profile to represent local structure. Scale models are more compact, have more information in the region close to boundary, and have less information in the region far away from the boundary. Therefore, a scale can be considered as a weighting factor. Experimental results are comparable to those of manual tracing by the radiologist and 2-D ASMs, and demonstrate that this algorithm is effective for the semi-automatic segmentation method of livers