Segmentation of computerized tomography images for three dimensional analysis

Accurate segmentation methods for the three-dimensional information contained in stacks of two-dimensional CT slices are required for object definition and solid modeling of objects within the data. Displays showing the surfaces of solid models present three-dimensional structural relationships more clearly than sequential viewing of the stack of slices. Methods of segmentation and their performance in the presence of noise typically found in CT images are considered with respect to accurate solid modeling, volume determination, and display quality. The same techniques can also be applied to general non-invasive examination procedures which generate three-dimensional density information representing internal objects.