Three-dimensional volume data collected by medical imaging techniques such as computed tomography (CT), magnetic resonance imaging (MRI) and single-photon emission computed tomography (SPECT) provides valuable information for many medical applications. However, such data is represented as a stack of parallel two-dimensional slices from which the anatomy is not easy to be visualized. Visualization techniques such as volume rendering run very slowly because vast three-dimensional arrays of voxels are collected in each scan and all voxels participate in the generation of each image. Existing isosurface extraction algorithms generate the surface as a triangular mesh with a large number of triangles which lead to long rendering time. The proposed algorithm extracts the isosurface in volume d...[ Read more ]

Three-dimensional volume data collected by medical imaging techniques such as computed tomography (CT), magnetic resonance imaging (MRI) and single-photon emission computed tomography (SPECT) provides valuable information for many medical applications. However, such data is represented as a stack of parallel two-dimensional slices from which the anatomy is not easy to be visualized. Visualization techniques such as volume rendering run very slowly because vast three-dimensional arrays of voxels are collected in each scan and all voxels participate in the generation of each image. Existing isosurface extraction algorithms generate the surface as a triangular mesh with a large number of triangles which lead to long rendering time. The proposed algorithm extracts the isosurface in volume data using a marching method in order to generate a smoother triangular mesh with a considerably smaller number of triangles. Moreover, extensions are proposed in varying the size of triangles adaptively to the local geometry of the surface to further reduce the number of triangles generated while preserving the resulting image quality. The proposed algorithms have been tested on synthetic as well as real data sets.