Algorithm for defining skeletal structures in biomedical models

Description of mechanical behavior of hard tissues by means of discrete models goes through various stages of analysis, which range from digital image processing to the specification of physical properties of tissue to the discrete model. This requires taking into account a key element: the decomposition of the model into its constituent parts. We conducted a bibliographic study of existing proposals for such decomposition, leading to the conclusion of the absence of a single strategy. There are several generic proposals, but these proved not to give a valid solution applicable to the cases examined corresponding to the articulations of the knee, hip and shoulder. In this paper we propose an algorithm to perform this decomposition by analyzing the spatial relationships between the contours present in consecutive planes. It is based on four stages: reading computer tomography (CT) slices, determining the contours that define bone tissue present on each slice, grouping of contours whose relationship meets a given criterion, and eliminating non-valid volumes. Results were compared with those obtained by means of Visualization ToolKit (VTK) and pyFormex, widely used in the visualization and analysis of medical imaging and modeling three-dimensional structures. As a main result, proposed algorithm under the same conditions and short processing time performs a better decomposition of anatomical models than the one made by VTK and pyFormex, with about a 90% of confidence