Accurate Topology and Geometry Reconstruction of Human Heads

An accurate topology and geometry structure of the human head is critical for numerical simulations and calculations. Automated reconstruction of multiple domains and watertight models based on medical images remains challenging. This paper improves the Moore boundary-tracing algorithm to achieve compact contour extraction. A region-contour relationship graph is proposed to infer a region's outer and inner contours in the radial direction of the image. A region-region connection graph is employed to express region correspondence in the axial direction. The region's one-to-many and many-to-many issues are all converted into one-to-one problems using the region decoupling method. Surface representation adopts free-form B-spline surfaces within a given accuracy. The Bspline technique fits the brain shape very well compared to triangular facets. An automatic assembly strategy is generated using the region-contour relationship and region-region connection graphs. The model's topology is exactly expressed by applying graph theory, and the geometric accuracy is controlled by the precision of user input. The innovation of this paper is establishing an accurate topological and geometrical human head model, which will significantly impact biological numerical calculation.