The objective of this thesis is to provide an improved characterization of porous scaffolds. A more focused objective is to provide a computational model simulating the cell mediated resorption process of resorbable bone substitutes. The thesis is structured in three scientific manuscripts. The first manuscript used fuzzy-based image treatment methods to analyse images generated by micro-computed tomography. From the literature, it is known that the fuzzy-based method helps to improve the accuracy of the characterization, in particular for scaffolds featuring a relatively small pore size. In addition, a new algorithm was introduced to determine both pore and interconnection sizes. The surface area of bone substitutes was quantified by using marching cube algorithm. Besides, the so-called Lattice Boltzmann method was used to characterize the permeability of the investigated scaffolds. Scaffolds made of [béta]-tricalcium phosphate ([béta]-Ca[subscript 3](PO[subscript 4])[subscript 2]) and presenting a constant porosity and four variable pore sizes were examined. The average pore size (diameter) of the four bone substitute groups (denominated with a letter from group A to D) was measured to be 170.3«1.7, 217.3«5.2, 415.8«18.8 and 972.3«10.9 [micro]m. Despite this significant change in pore size, the pore interconnection size only increased slightly, in the range of 61.7 to 85.2 [micro]m. The average porosity of the four groups was 52.3«1.5 %. The surface density of scaffolds decreased from 11.5 to 3.3 mm[superscript -1], when the pore size increased. The results revealed that the permeability of scaffolds is in the same order of magnitude and increased from 1.1?10[superscript -10] to 4.1?10[superscript -10] m[superscript 2] with increasing the pore size. The second manuscript was devoted to the use of subvoxelization algorithm and high-resolution scanner, in an attempt to further improve the accuracy of the results, in particular, of the small pore scaffolds. As expected, an increase of the image resolution from 15 to 7.5 [micro]m significantly eased the segmentation process and hence improved scaffold characterization. Subvoxelization also improved the results specifically in terms of interconnection sizes. Specifically, much smaller interconnection sizes were yielded after applying the subvoxelization process. For example, the mean interconnection size of small pore size groups, group A and B, dropped from 63 to 20 and 30 [micro]m, respectively. Furthermore, due to more details obtained from subvoxelization and high-resolution scanning, additional effects so called"boundary effects" were observed. The boundary effects can yield misleading results in terms of interconnection sizes. The means to reduce these effects were proposed. The third manuscript focused on the simulation and understanding of cell mediated resorption of bone graft substitutes. A computer model was developed to simulate the resorption process of four bone substitute groups. [mu]CT data and new"image processing" tools such as labelling and skeletonization were combined in an algorithm to perform the steps of resorption simulation algorithm. The proposed algorithm was verified by comparing simulation results with the analytical results of a simple geometry and biological in vivo data of bone substitutes. A correlation coefficient between the simulation results and both analytical and experimental data, was found to be larger than 0.9. Local resorption process revealed faster resorption in external region specifically at earlier resorption time. This finding is in agreement with the in vivo results. Two definitions were introduced to estimate the resorption rate; volume resorption rate and linear resorption rate. The volume resorption rate was proportional to accessible surface and decreased when the pore size increased, while the linear resorption rate was proportional to thickness of material and increased with increasing the pore size. In addition, the simulation results revealed no effect of resorption direction on resorption behaviour of substitutes. However, the resorption rate of small pore size samples was decreased with increasing the minimum interconnection size required for cell ingrowth, to 100 [micro]m. This thesis combined novel"image processing" tools and subvoxelization method to improve the characterization of porous bone substitutes used in the bone repair process. The improved characterization allowed a more accurate simulation process. The simulation data were consistent with previously obtained biological data of the same group and allows understanding the local resorption process. The available tools and results are expected to help with the design of optimal substitute for bone repair."--Résumé abrégé par UMI.