“A tool for fitting surfaces to medical image data” by Sowell and Grimm

MRI and CT scanners have long been used to produce three-dimensional samplings of anatomy elements for use in medical visualization and analysis. From such datasets, it is often desired that a smooth, parameterized surface be reconstructed. Such models are useful not only for visualization, but also as finite element models on which simulations and further processing can be performed. Many algorithms have been proposed for reconstructing surfaces from a series of contours. [Keppel 1975] and [Fuchs et al. 1977]  proposed the first algorithms for “contour stitching”. Many improvements have been made since then, but the general approach remains the same: reconstruct a surface by connecting the vertices of adjacent contours in order to generate a mesh that passes through all contours. These methods all assume that a series of parallel contours on the input data are given. Unfortunately, the act of marking these contours is currently a time-consuming problem involving a great deal of manual intervention. In this process, an experienced scientist or physician must manually segment the original volume dataset by going through two-dimensional slices of the data one by one, and marking a series of contours that outline the object of interest.  

1. Fuchs, H., Kedem, Z. M., and Uselton, S. P. 1977. Optimal surface reconstruction from planar contours. Commun. ACM 20, 10, 693–702.
2. Grimm, C., Laidlaw, D., and Crisco, J. 2002. Fitting Manifold Surfaces To 3D Point Clouds. Journal of Biomechanical Engineering, 124, 136–140.
3. Keppel, E. 1975. Approximating complex surfaces by triangulation of contour lines. IBM Journal of Research and Development, 10, 2–11.