Processing of Vertebral CT-Images for Diagnosis of Osteoporosis

P. Saparin
Institut für Physik, Universität Potsdam

Since in any object mass and structure depend on each other to build a functional form, methodologies to measure structural information are needed to understand the contribution of structure to support the form, the complexity of the spatial architecture, the changes and the loss of object integrity. In the past, the structure of hierarchical natural composites, especially biomaterials, has been evaluated merely by parameters measuring parts of architectural elements such as thickness, number of elements, the space in between, the volume, or the surface of these elements. They are all based on invasive techniques.

We propose a non-invasive technique to assess structure in its complex spatial distribution by describing and quantifying the structural architecture as a whole. Vertebral computed tomography (CT) images were preprocessed by newly developed algorithms which separate the vertebra from the soft tissue and then split the entire vertebra into the trabecular bone and cortical shell. Next the concept of measures of complexity based on symbolic dynamics is applied to segmented CT-images which are, crucially important, symbol-encoded by both static and dynamical approaches. To study their structural properties, we generalize the concept of symbolic dynamics into two-dimensional case.

This method is employed to identify structural changes in human cancellous bone of vertebral bodies and leads to new insides for the understanding of bone's internal structure. The results give the first experimental and quantitative evidence that the complexity of cancellous bone structure declines rapidly with increased disintegration of bone and exponentially relates to the amount of material it is built from. In addition, this method is significantly sensitive to changes in structure and provides improvements of differentiation of structural loss and may well have an impact beyond biological and physical science, expending materials science such as metallurgy, bioengineering, and mechanical engineering.