The routine use of 3D imaging methods by the medical community, such as MRI and CT is creating a vast amount of volumetric data. The usual way of storing these data is voxel images that allow visual capturing of the objects, but lack structural description since they are based on huge lists of numbers.On the other hand new medical imaging modalities, such as Optical Tomography require precise and smooth models of human tissues (e.g. babyís head) for the solution of the inverse and forward problem. Until now the models used in such applications are rough polygon meshes based on voxel images. We demonstrate a method to create a parametric model of a body structure by mapping the surface to a sphere. Basis functions like spherical harmonics can be used for a description of the surface.

Parametric Description of 3-Dimensional Surfaces
Spherical Harmonics
The closed surface extracted from a babyís head voxel image is topologically equivalent to a sphere, therefore it can be mapped on a sphereís surface. The construction of this homeomorphism from the closed surface to the unit sphere results in a parametric representation of the surface. Diffusion of the surface nodes is used for an initial mapping and then constrained non-linear optimisation method to achieve the necessary one-to-one mapping between the two surfaces. The square faces from the voxel object have been mapped on the sphereís surface. Two parameters Theta and Fi have been assigned for each node of the headís surface. For every point on the headís surface a unique pair of parameters has been assigned.
Using weighted basis functions also described on the sphere, such as the Spherical Harmonics we can reconstruct the shape of the voxel object to any desired level of detail
Having defined the Spherical Harmonics weighting coefficients [Cx, Cy , Cz], the surface is now described, by the parametric function f (theta,fi):
>>Mesh Creation<<