ABSTRACT
The main aim of bone remodeling is to repair damaged bone. During normal activity, microdamage occurs in the form of linear microcracks and diffuse damage which appear in response to trabecular stress induced by external loading. Recent research concerns the study of microdamage paying attention to its localizations and influence of local trabecular microarchitecture. After initiation, enhancement or attenuation of a linear microcrack might depend on its location. Correlation between microcrack density and structure model index (SMI) or with rods in a particular direction was shown in (Shi, Liu, Wang, Guo, & Niebur 2010). An improvement of our remodeling process should incorporate damage formation. Microdamage could be introduced directly in the bone matrix state or by adding new constraints
1 INTRODUCTION
In a previous study (Mellouli & Ricordeau 2011) we presented a stochastic process for simulating trabecular bone surface remodeling. This kind of bone is a porous tissue made up of a mesh-like network of tiny pieces of bone called trabeculae. During remodeling, cellularactivity takes place on bone surface in what is called Bone Multicellular Units (BMU). The process which was proposed for simulations is an iterative stochastic germ-grain model. It is therefor a BMU-model where germ corresponds to BMU origination and grain corresponds to BMU resorption shape. Driven by a limited number of parameters that match biological ones, such a model can be considered a black-box model. Three main global parameters relating to biological ones are used to implement this temporal process. Intensity parameter λ relates to the activation frequency, hence to the number of BMUs originated at each process iteration. Proportion α of pixels to be resorbed relates to bone imbalance, hence the life-span and number of resorption cells called osteoblasts. In a normal context, bone remodeling is a targeted biological process which ensures that old bone is renewed. High value of κ relates to a high targeted process. κ is used in a constraint introduced to validate each new proposed BMU, compared to the local state of the bone matrix where voxel-values are simply expressed as the age from last formation (number of iterations). In this process, λ, α and κ are global parameters. Type and orientation of local structures (plate-like and rod-like) are not taken into account. Such an isotropic process has been implemented using realistic three-dimensional binary images for initialization. Such images correspond to 3D-binarized bone samples V1 and V2 (figures 1,2) which were
used to validate each BMU location and shape. Hence local information on the structure could be incorporated such as type of local structure and its orientation.
