The next step is to take that information from the visualization stage to the simulation stage.  One limitation in the simulation process is that the Visible Human specimens represent individual subjects.  However, researchers would like to be able to simulate a wide range of subjects based on age, weight, height, and sex. Thus, it would be valuable to be able to take a specific 3D anatomical model and through computational morphology, transform that person to any other person. The Visible Human data set would then not only provide information of a single specimen, but through anthropometric scaling, could represent a wide range of body types, potentially at various chronological development stages.  In this paper, we present our preliminary work in deforming the full body anatomy database to different body types using anthropometric information from other subjects.

      Much work has been done in the area of modeling structural and dynamic attributes related to the Visible Human data ([3]).  We focus here on the deformation of the musculoskeletal anatomy from the Visible Human male data set.  Typically, body modeling and deformation has used metaball simulation of body parts based on age, gender, and race ([4]).  This approach shows the potential for applying these same parameters to the Visible Human data set to achieve anatomically accurate models of a virtually unlimited number of body forms.

      Our initial experiment has been to take multiple external measurements of the alternate body type and generate an external lattice structure from those dimensions. The external lattice structure for each body region is calculated based on the bounding box dimensions of each model from the 3D anatomy database, in the particular region of interest.  Each model is transformed based on its tissue type. In this case, we have chosen the fraction of the scaling associated with hard tissue to be less than that of soft tissue. Figure 1 shows the original musculoskeletal models from the Visible Human male’s right lower extremity. Figure 2 shows the models after being transformed to a manually calculated set of anthropological dimensions.

      The results show excellent promise, but are limited in two respects.  First, based solely on a limited set of external measurements, the scaling of hard tissue appears to be large, particularly at the heads of the long bones.  This we feel can be resolved once we have access to a database of skeletal dimensions that will allow us to deform each bone to a specific set of scale factors.  The second limitation involves the weighted scaling of soft tissue, particularly muscle.  In our initial efforts, the weighting of each vertex was based on the mass and center of mass of each muscle.  Again, we believe that distortions can be alleviated by applying additional levels of information about muscle dimensionality from a  variety of different body types.

      In earlier work, we have described the use of incorporating acceleration of muscle mass relative to the acceleration of body segments and detailed the key steps in achieving this end: construct surface anatomy; derive parameters; apply muscle morphology; incorporate motion equations ([5]).  Considerable work has been done in the encapsulation of critical information, such as structure, topology, and the mechanics of the model.  Using this information, physical simulation and motion-based deformation of the 3D reconstructed data can be performed.  By expanding the preliminary work presented here, accurate morphological deformation of the Visible Human male 3D database to other topological forms can be achieved.  Subsequently, it is hoped that kinematic information can also be quantitatively translated to a large number of anatomically accurate human body models.