The influence of image processing of -CT images on mechanical behavior of mandibular trabecular bone structure using micro finite element method

ScienceDirect Available online at www.sciencedirect.com Procedia Structural Integrity 43 (2023) 276–281 2452-3216 © 2023 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under the responsibility of MSMF10 organizers. 10.1016/j.prostr.2022.12.271 10.1016/j.prostr.2022.12.271 2452-3216 © 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license ( https://creativecommons.org/licenses/by-nc-nd/4.0 ) Peer-review under the responsibility of MSMF10 organizers. Available online at www.sciencedirect.com ScienceDirect Structural Integrity Procedia 00 (2022) 000–000 www.elsevier.com/locate/procedia 2452-3216 © 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under the responsibility of MSMF10 organizers. 10th International Conference on Materials Structure and Micromechanics of Fracture The influence of image processing of μ-CT images on mechanical behavior of mandibular trabecular bone structure using micro finite element method Apolena Šustkováa*, Barbora Thomkováa, Tomáš Zikmundb, Jozef Kaiserb, Marek Joukalc, Petr Marciána aFaculty of Mechanical Engineering, Brno University of Technology, Technická 2896/2, Brno, 616 69, Czech Republic bCEITEC – Central European Institute of Technology, Brno University of Technology, Purkyňova 656/123, Brno, 612 00, Czech Republic cDepartment of Anatomy, Faculty of Medicine, Masaryk University, Kamenice 753/5, Brno, 625 00, Czech Republic Abstract Computational modeling plays an important role in assessment of biomaterial structures, living tissue and their mechanical properties. Microstructure investigation of bone tissue needs high-level computational models based on μ-CT images. A crucial factor when creating a computational model is image processing using tissue segmentation. The most common method for segmentation is thresholding. Threshold value can influence the geometry model and mechanical response in the bone tissue. One of the aims of the work is to determine and analyze the apparent mechanical behavior of mandibular trabecular bone structure based on different threshold value. Another aim of this study is to assess its influence on strain intensity from a local point of view, namely on one specific trabeculae. Fifty-four computational models including detailed trabecular structure of mandible were created based on μ-CT images. The thresholding was observed to have a significant effect on mechanical behavior of the trabecular bone structure. © 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under the responsibility of MSMF10 organizers. Keywords: image processing; trabecular bone structure; thresholding; µ-CT; FEM * Corresponding author. E-mail address: [email protected] Available online at www.sciencedirect.com ScienceDirect Structural Integrity Procedia 00 (2022) 000–000 www.elsevier.com/locate/procedia 2452-3216 © 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under the responsibility of MSMF10 organizers. 10th International Conference on Materials Structure and Micromechanics of Fracture The influence of image processing of μ-CT images on mechanical behavior of mandibular trabecular bone structure using micro finite element method Apolena Šustkováa*, Barbora Thomkováa, Tomáš Zikmundb, Jozef Kaiserb, Marek Joukalc, Petr Marciána aFaculty of Mechanical Engineering, Brno University of Technology, Technická 2896/2, Brno, 616 69, Czech Republic bCEITEC – Central European Institute of Technology, Brno University of Technology, Purkyňova 656/123, Brno, 612 00, Czech Republic cDepartment of Anatomy, Faculty of Medicine, Masaryk University, Kamenice 753/5, Brno, 625 00, Czech Republic Abstract Computational modeling plays an important role in assessment of biomaterial structures, living tissue and their mechanical properties. Microstructure investigation of bone tissue needs high-level computational models based on μ-CT images. A crucial factor when creating a computational model is image processing using tissue segmentation. The most common method for segmentation is thresholding. Threshold value can influence the geometry model and mechanical response in the bone tissue. One of the aims of the work is to determine and analyze the apparent mechanical behavior of mandibular trabecular bone structure based on different threshold value. Another aim of this study is to assess its influence on strain intensity from a local point of view, namely on one specific trabeculae. Fifty-four computational models including detailed trabecular structure of mandible were created based on μ-CT images. The thresholding was observed to have a significant effect on mechanical behavior of the trabecular bone structure. © 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under the responsibility of MSMF10 organizers. Keywords: image processing; trabecular bone structure; thresholding; µ-CT; FEM * Corresponding author. E-mail address: [email protected] 2 Author name / Structural Integrity Procedia 00 (2022) 000–000 1. Introduction Computational modeling using the finite element method (FEM) is increasingly widespread in addressing the issue of mechanical interaction of dental implants with bone tissue. In this case, the computational model of the bone tissue of the mandible or maxilla must include the geometry of the trabecular micro architecture, which consists of rod and plate structures. The study and mechanical analysis of these bone structures is made possible by micro FEM models. For this purpose, it is necessary to use data obtained on a micro-computer tomography (μ-CT) device. The output of this device is high-resolution image data, which must be transformed into a computational model using image processing. Image segmentation is used for this purpose. Specifically, for micro CT data, the most often method used is thresholding (van Eijnatten et al. 2017). This process is crucial and will most affect the subsequent 3D reconstruction of the model of geometry of the computational model and, as a result, also the related deformation and stress states. Threshold determines the bone and soft tissue interface – pixels with values greater than the threshold value are included in the computational model and pixels with values under the threshold are considered unimportant (soft tissue, fat, etc.). Moreover, the threshold value for automatic and manual segmentation may differ and the variant used depends, among other things, on the software. All of this can have an impact on the final geometry model of bone tissues and thus on the computational model (which subsequently also affects stress-strain stages). The aim of this study is to determine and analyze apparent mechanical behavior of mandibular trabecular bone structure based on image processing using different threshold values and assessment of its influence on local strain intensity in chosen rod. 2. Materials and methods The mandibular bone segment (20 x 15 x 15 mm) was acquired from the Department of Anatomy, Faculty of Medicine, Masaryk University Brno, Czech Republic. This sample was scanned using a micro-computed tomography scanner (GE phoenix v|tome|x L240, GE Sensing & Inspection Technologies GmbH, Wunstorf, Germany) with a voxel size of 15 μm. The bone density calibration was performed using a hydroxyapatite phantom (MicroCT-HA D20, © QRM GmbH, Moehrendorf, Germany) included during scanning. This phantom caliber consists of 5 cylinders with different densities – 0 g/cm3, 0.05 g/cm3, 0.20 g/cm3, 0.80 g/cm3, 1.20 g/cm3 (see Fig. 1a)). Micro CT images of phantom were then used to determine bone density according to a standard procedure based on a linear relationship between hydroxyapatite density and the corresponding image – pixel gray values (see Fig. 1b)). Figure 1: Phantom caliber: a) Transversal view of the hydroxyapatite phantom block (g/cm3); b) Linear relationship between bone density of hydroxyapatite samples and pixel gray value. 2.1. Image processing and models of geometry For subsequent analysis, a region of interest (ROI) in the area of the alveolar bone in the size of 4 x 4 x 4 mm was selected (see Fig. 2a)). This ROI was chosen to represent a typical location for dental implant insertion. Automatic segmentation was performed in the ImageJ software (Schneider et al. 2012) for different threshold values (pixel gray values) corresponding to bone density from 0.25 to 0.65 g/cm3 (see Fig. 2b)) with the increment of 0.05 g/cm3 (nine 1.55 0 10 CT 1. 51 0. 0 0.2 0. 0. 0. 1 1.2 1. 05000 10000 15000 20000 Bone density ( g cm Pi el gray value ( T a) b) Apolena Šustková et al. / Procedia Structural Integrity 43 (2023) 276–281 277 Available online at www.sciencedirect.com ScienceDirect Structural Integrity Procedia 00 (2022) 000–000 www.elsevier.com/locate/procedia 2452-3216 © 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under the responsibility of MSMF10 organizers. 10th International Conference on Materials Structure and Micromechanics of Fracture The influence of image processing of μ-CT images on mechanical behavior of mandibular trabecular bone structure using micro finite element method Apolena Šustkováa*, Barbora Thomkováa, Tomáš Zikmundb, Jozef Kaiserb, Marek Joukalc, Petr Marciána aFaculty of Mechanical Engineering, Brno University of Technology, Technická 2896/2, Brno, 616 69, Czech Republic bCEITEC – Central European Institute of Technology, Brno University of Technology, Purkyňova 656/123, Brno, 612 00, Czech Republic cDepartment of Anatomy, Faculty of Medicine, Masaryk University, Kamenice 753/5, Brno, 625 00, Czech Republic Abstract Computational modeling plays an important role in assessment of biomaterial structures, living tissue and their mechanical properties. Microstructure investigation of bone tissue needs high-level computational models based on μ-CT images. A crucial factor when creating a computational model is image processing using tissue segmentation. The most common method for segmentation is thresholding. Threshold value can influence the geometry model and mechanical response in the bone tissue. One of the aims of the work is to determine and analyze the apparent mechanical behavior of mandibular trabecular bone structure based on different threshold value. Another aim of this study is to assess its influence on strain intensity from a local point of view, namely on one specific trabeculae. Fifty-four computational models including detailed trabecular structure of mandible were created based on μ-CT images. The thresholding was observed to have a significant effect on mechanical behavior of the trabecular bone structure. © 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under the responsibility of MSMF10 organizers. Keywords: image processing; trabecular bone structure; thresholding; µ-CT; FEM * Corresponding author. E-mail address: [email protected] Available online at www.sciencedirect.com ScienceDirect Structural Integrity Procedia 00 (2022) 000–000 www.elsevier.com/locate/procedia 2452-3216 © 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under the responsibility of MSMF10 organizers. 10th International Conference on Materials Structure and Micromechanics of Fracture The influence of image processing of μ-CT images on mechanical behavior of mandibular trabecular bone structure using micro finite element method Apolena Šustkováa*, Barbora Thomkováa, Tomáš Zikmundb, Jozef Kaiserb, Marek Joukalc, Petr Marciána aFaculty of Mechanical Engineering, Brno University of Technology, Technická 2896/2, Brno, 616 69, Czech Republic bCEITEC – Central European Institute of Technology, Brno University of Technology, Purkyňova 656/123, Brno, 612 00, Czech Republic cDepartment of Anatomy, Faculty of Medicine, Masaryk University, Kamenice 753/5, Brno, 625 00, Czech Republic Abstract Computational modeling plays an important role in assessment of biomaterial structures, living tissue and their mechanical properties. Microstructure investigation of bone tissue needs high-level computational models based on μ-CT images. A crucial factor when creating a computational model is image processing using tissue segmentation. The most common method for segmentation is thresholding. Threshold value can influence the geometry model and mechanical response in the bone tissue. One of the aims of the work is to determine and analyze the apparent mechanical behavior of mandibular trabecular bone structure based on different threshold value. Another aim of this study is to assess its influence on strain intensity from a local point of view, namely on one specific trabeculae. Fifty-four computational models including detailed trabecular structure of mandible were created based on μ-CT images. The thresholding was observed to have a significant effect on mechanical behavior of the trabecular bone structure. © 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under the responsibility of MSMF10 organizers. Keywords: image processing; trabecular bone structure; thresholding; µ-CT; FEM * Corresponding author. E-mail address: [email protected] 2 Author name / Structural Integrity Procedia 00 (2022) 000–000 1. Introduction Computational modeling using the finite element method (FEM) is increasingly widespread in addressing the issue of mechanical interaction of dental implants with bone tissue. In this case, the computational model of the bone tissue of the mandible or maxilla must include the geometry of the trabecular micro architecture, which consists of rod and plate structures. The study and mechanical analysis of these bone structures is made possible by micro FEM models. For this purpose, it is necessary to use data obtained on a micro-computer tomography (μ-CT) device. The output of this device is high-resolution image data, which must be transformed into a computational model using image processing. Image segmentation is used for this purpose. Specifically, for micro CT data, the most often method used is thresholding (van Eijnatten et al. 2017). This process is crucial and will most affect the subsequent 3D reconstruction of the model of geometry of the computational model and, as a result, also the related deformation and stress states. Threshold determines the bone and soft tissue interface – pixels with values greater than the threshold value are included in the computational model and pixels with values under the threshold are considered unimportant (soft tissue, fat, etc.). Moreover, the threshold value for automatic and manual segmentation may differ and the variant used depends, among other things, on the software. All of this can have an impact on the final geometry model of bone tissues and thus on the computational model (which subsequently also affects stress-strain stages). The aim of this study is to determine and analyze apparent mechanical behavior of mandibular trabecular bone structure based on image processing using different threshold values and assessment of its influence on local strain intensity in chosen rod. 2. Materials and methods The mandibular bone segment (20 x 15 x 15 mm) was acquired from the Department of Anatomy, Faculty of Medicine, Masaryk University Brno, Czech Republic. This sample was scanned using a micro-computed tomography scanner (GE phoenix v|tome|x L240, GE Sensing & Inspection Technologies GmbH, Wunstorf, Germany) with a voxel size of 15 μm. The bone density calibration was performed using a hydroxyapatite phantom (MicroCT-HA D20, © QRM GmbH, Moehrendorf, Germany) included during scanning. This phantom caliber consists of 5 cylinders with different densities – 0 g/cm3, 0.05 g/cm3, 0.20 g/cm3, 0.80 g/cm3, 1.20 g/cm3 (see Fig. 1a)). Micro CT images of phantom were then used to determine bone density according to a standard procedure based on a linear relationship between hydroxyapatite density and the corresponding image – pixel gray values (see Fig. 1b)). Figure 1: Phantom caliber: a) Transversal view of the hydroxyapatite phantom block (g/cm3); b) Linear relationship between bone density of hydroxyapatite samples and pixel gray value. 2.1. Image processing and models of geometry For subsequent analysis, a region of interest (ROI) in the area of the alveolar bone in the size of 4 x 4 x 4 mm was selected (see Fig. 2a)). This ROI was chosen to represent a typical location for dental implant insertion. Automatic segmentation was performed in the ImageJ software (Schneider et al. 2012) for different threshold values (pixel gray values) corresponding to bone density from 0.25 to 0.65 g/cm3 (see Fig. 2b)) with the increment of 0.05 g/cm3 (nine 1.55 0 10 CT 1. 51 0. 0 0.2 0. 0. 0. 1 1.2 1. 05000 10000 15000 20000 Bone density ( g cm Pi el gray value ( T a) b) 278 Apolena Šustková et al. / Procedia Structural Integrity 43 (2023) 276–281 Author name / Structural Integrity Procedia 00 (2022) 000–000 3 segmented datasets). Nine geometries were then created in STL Model Creator (Marcián et al. 2011 programmed in Matlab 2011b (Math Works, Natick MA, USA). Figure 2c) shows geometries with thresholds 0.25 and 0.65 g/cm3. The effect of the threshold is evident in the detail of the figure – as the value of the threshold increases, the amount of bone tissue decreases. The difference between the volumes of these two geometries yields 38%. Figure 2: Considered ROI: a) Selected micro-CT image with highlighted ROI; b) Comparison of one image segmented with different threshold values (0.25 g/cm3 up vs 0.65g/cm3 down); c) Comparison of different threshold values for ROI (0.25 g/cm3 on the left vs 0.65g/cm3 on the right). The circles indicate in more detail how threshold value affects the amount of bone tissue. Orientation of coordinate system: x-labiolingual, y-mesiodistal, z-occlusal. 2.2. FE mesh ANSYS® Academic esearch Mechanical, elease 1 .0 (Swanson Analysis, Inc., Houston, PA, USA was used to create the micro FEM models as well as FE calculation. Each geometry model was discretized using volume quadratic element SOLID 187 with the global element size 0.03 mm. The number of nodes ranged from 5467559 to 8453260. 2.3. Material model Two variants of material models were used. The first material model was assumed as linear, isotropic, and homogeneous with material properties of Young's modulus 5000 MPa (Costa et al. 2017; Rahmoun et al. 2014) and Poisson's ratio 0.3 (Kayabaşı et al. 200 ; Natali et al. 200 . It is the linear homogeneous model that is most often used and sufficient for most analyzes (Viceconti 2011). The second material model was assumed as nonhomogeneous based on micro CT data (Marcián et al. 2021 . Local values of the Young's modulus are determined based on the density obtained from the pixel gray value according to the following relationship (1) (Inagawa et al. 2018): 𝐸𝐸 = 17486 ∙ 𝜌𝜌1.596 (1) where E is Young's modulus MPa and ρ is density g cm3] given by the following equation: 𝜌𝜌 = 1.5570 ∙10−4 ∙𝐶𝐶𝐶𝐶 − 1.4519 (2) a) b) c) 4 Author name / Structural Integrity Procedia 00 (2022) 000–000 In Eq. (2), CT is a unitless gray value (pixel intensity) obtained from the image datasets with calibration phantom, see Fig. 1 b). For the models with nonhomogeneous distribution, Young's modulus was calculated for each pixel of all considered datasets and mapped onto the corresponding finite element mesh. Using the software CTPixelMapper (Borák and Marcián 201 , a script was generated, which was used to assign the material properties to the geometry of the trabecular bone (see Fig. 3). Figure 3: Nonhomogeneous distribution of Young’s modulus in the trabecular structure for variants 0.25N and 0.65N. 2.4. Load and boundary conditions To determine the apparent material properties in three considered directions, the loads were applied on the model based on the literature (Ševeček et al. 201 a, 201 b . Loads were applied on a layer of nodes of 0.05 mm thickness. Regarding the boundary conditions, fixed support was applied to one surface preventing movement in all directions. On the opposite surface, the remote displacement was applied (set as rigid). Remote displacement was set as 1% of the length of the segment. 2.5. Model variants From one dataset, nine geometry models were created, differing with the choice of a threshold value according to the density marked as 0.25; 0.3; 0.35; 0.4; 0.45; 0.5; 0.55; 0.6; 0.65. In addition, two material models were used for each of these models – homogeneous and nonhomogeneous (further referred to as H and N). Each of these models was loaded in three directions. Thus, a total of 54 model variants (0.25H, 0.3H, …, 0. 5H, 0.25N, …, 0.65N) were investigated. 3. Results The analysis of apparent Young's modulus (Eapp) in each direction was performed. The value of Eapp was calculated in three directions (x, y, z) according to (Ševeček et al. 201 a, 201 b . In Figure 4, all determined values of Eapp are shown. Furthermore, the distribution of strain intensity (defined as maximum shearing strain) in one particular trabeculae was analyzed. Figure 5 presents the results for the variant with loading in the Y axis direction for all solved variants of the models of geometry. Strain isolines are plotted with the same color range. In addition, the results are plotted in the graph using a path on the same trabeculae (variants of models of geometry with minimum and maximum threshold values are selected for this comparison). Apolena Šustková et al. / Procedia Structural Integrity 43 (2023) 276–281 279 Author name / Structural Integrity Procedia 00 (2022) 000–000 3 segmented datasets). Nine geometries were then created in STL Model Creator (Marcián et al. 2011 programmed in Matlab 2011b (Math Works, Natick MA, USA). Figure 2c) shows geometries with thresholds 0.25 and 0.65 g/cm3. The effect of the threshold is evident in the detail of the figure – as the value of the threshold increases, the amount of bone tissue decreases. The difference between the volumes of these two geometries yields 38%. Figure 2: Considered ROI: a) Selected micro-CT image with highlighted ROI; b) Comparison of one image segmented with different threshold values (0.25 g/cm3 up vs 0.65g/cm3 down); c) Comparison of different threshold values for ROI (0.25 g/cm3 on the left vs 0.65g/cm3 on the right). The circles indicate in more detail how threshold value affects the amount of bone tissue. Orientation of coordinate system: x-labiolingual, y-mesiodistal, z-occlusal. 2.2. FE mesh ANSYS® Academic esearch Mechanical, elease 1 .0 (Swanson Analysis, Inc., Houston, PA, USA was used to create the micro FEM models as well as FE calculation. Each geometry model was discretized using volume quadratic element SOLID 187 with the global element size 0.03 mm. The number of nodes ranged from 5467559 to 8453260. 2.3. Material model Two variants of material models were used. The first material model was assumed as linear, isotropic, and homogeneous with material properties of Young's modulus 5000 MPa (Costa et al. 2017; Rahmoun et al. 2014) and Poisson's ratio 0.3 (Kayabaşı et al. 200 ; Natali et al. 200 . It is the linear homogeneous model that is most often used and sufficient for most analyzes (Viceconti 2011). The second material model was assumed as nonhomogeneous based on micro CT data (Marcián et al. 2021 . Local values of the Young's modulus are determined based on the density obtained from the pixel gray value according to the following relationship (1) (Inagawa et al. 2018): 𝐸𝐸 = 17486 ∙ 𝜌𝜌1.596 (1) where E is Young's modulus MPa and ρ is density g cm3] given by the following equation: 𝜌𝜌 = 1.5570 ∙10−4 ∙𝐶𝐶𝐶𝐶 − 1.4519 (2) a) b) c) 4 Author name / Structural Integrity Procedia 00 (2022) 000–000 In Eq. (2), CT is a unitless gray value (pixel intensity) obtained from the image datasets with calibration phantom, see Fig. 1 b). For the models with nonhomogeneous distribution, Young's modulus was calculated for each pixel of all considered datasets and mapped onto the corresponding finite element mesh. Using the software CTPixelMapper (Borák and Marcián 201 , a script was generated, which was used to assign the material properties to the geometry of the trabecular bone (see Fig. 3). Figure 3: Nonhomogeneous distribution of Young’s modulus in the trabecular structure for variants 0.25N and 0.65N. 2.4. Load and boundary conditions To determine the apparent material properties in three considered directions, the loads were applied on the model based on the literature (Ševeček et al. 201 a, 201 b . Loads were applied on a layer of nodes of 0.05 mm thickness. Regarding the boundary conditions, fixed support was applied to one surface preventing movement in all directions. On the opposite surface, the remote displacement was applied (set as rigid). Remote displacement was set as 1% of the length of the segment. 2.5. Model variants From one dataset, nine geometry models were created, differing with the choice of a threshold value according to the density marked as 0.25; 0.3; 0.35; 0.4; 0.45; 0.5; 0.55; 0.6; 0.65. In addition, two material models were used for each of these models – homogeneous and nonhomogeneous (further referred to as H and N). Each of these models was loaded in three directions. Thus, a total of 54 model variants (0.25H, 0.3H, …, 0. 5H, 0.25N, …, 0.65N) were investigated. 3. Results The analysis of apparent Young's modulus (Eapp) in each direction was performed. The value of Eapp was calculated in three directions (x, y, z) according to (Ševeček et al. 201 a, 201 b . In Figure 4, all determined values of Eapp are shown. Furthermore, the distribution of strain intensity (defined as maximum shearing strain) in one particular trabeculae was analyzed. Figure 5 presents the results for the variant with loading in the Y axis direction for all solved variants of the models of geometry. Strain isolines are plotted with the same color range. In addition, the results are plotted in the graph using a path on the same trabeculae (variants of models of geometry with minimum and maximum threshold values are selected for this comparison). 280 Apolena Šustková et al. / Procedia Structural Integrity 43 (2023) 276–281 Author name / Structural Integrity Procedia 00 (2022) 000–000 5 Figure 4: Values of Eapp in all directions (x,y,z – same as in Figure 2c)) for the homogeneous (on the left) and nonhomogeneous (on the right) model variants. Figure 5: Strain intensity in considered trabeculae (on the left) and along the path which was created based on the same start and end point (on the right). 4. Discussion Thresholding is a basic image processing tool for creating computational models from μ-CT images. The choice of the threshold fundamentally affects the results of stress-strain stages. As confirmed by the results of this study, the difference in the estimation of apparent mechanical properties is 74 % for homogeneous material and up to 59% for nonhomogeneous material when comparing the variants of models of geometry with minimum and maximum threshold values (0.25 and 0.65). In the case of a homogeneous material, the decrease in Eapp is linear and corresponds to the decrease in bone volume, which has the same value of E = 5000 MPa everywhere. When considering nonhomogeneous material, the decrease in Eapp with the decreasing volume for variants 0.25N, 0.30N and 0.35N is not so significant. This is due to the fact that the bone tissue at the edges does not have such a high density and its E, given by relation (1), is not so high, so there will be no significant decrease in Eapp as in the case of a homogeneous material. For other variants of nonhomogeneous material, the decrease in Eapp is more pronounced. Eapp values are relatively low for both variants of material, as it is a bone tissue sample with high porosity (83% to 90%). This corresponds to very low-density bone tissue (classification D4 according to Misch 2008). Nevertheless, values of Eapp obtained in this work correspond to the observations of the literature (van Eijden et al. 200 ; Marcián et al. 201 . From the comparison of the local stress of observed trabeculae, it is evident that the choice of the threshold has a more significant effect on the character of strain intensity distribution than the used material model. Character of the distribution for both selected material models is approximately the same. From the course of the strain intensity in the graph in Fig. 5 it can be seen that due to the change of the threshold value there is a substantial redistribution of strain. The difference of strain intensity between the models 0.25H and 0.65H is 55% and between the models 0.25N and 0.65N is 43%. 6 Author name / Structural Integrity Procedia 00 (2022) 000–000 5. Conclusion Using data from micro CT and image processing using automatic segmentation by thresholding, computational models were created and a comparative analysis was performed using micro FEM. The main observations of this study may be summarized as follows: 1. It is evident that the change of the threshold (the increasing value of threshold) has an impact on the amount of bone tissue. In this study, the difference in volumes between geometry models 0.25 and 0.65 is up to 38%. 2. The value of Eapp is decreasing with the increasing value of the threshold and thus with decreasing amount of the bone tissue for both material models. 3. Although the difference between the material used in this study is not so significant, the values of strain intensity and its distribution is dependent on the chosen threshold. Acknowledgement The work was supported by projects FSI-S-20-6164 and FSI-S-20-6175. References Borák, L., Marcián, P., 2017. Inhomogeneous Material Properties Assignment to Finite Element Models of Bone: A Sensitivity Study. Engineering Mechanics 2017, 190–193. osta, M. ., Tozzi, G., ristofolini, L., Danesi, V., Viceconti, M., Dall’Ara, E., 2017. Micro Finite Element models of the vertebral body: Validation of local displacement predictions. PLOS ONE 12(7), e0180151. Furukawa, T., Matsunaga, S., Morioka, T., Nakano, T., Abe, S., Yoshinari, M., Yajima, Y., 2019. Study on bone quality in the human mandibleAlignment of biological apatite crystallites. Journal of Biomedical Materials Research Part B: Applied Biomaterials 107(3), 838–846. Inagawa, H., Suzuki, N., Aoki, K., Wakabayashi, N., 201 . Potential for estimation of Young’s modulus based on computed tomography numbers in bone: A validation study using a nano-indentation test on murine maxilla. Dental, Oral and Craniofacial Research 4(4), 1-7. Kayabaşı, O., Yüzbasıoğlu, E., Erzincanlı, F., 2006. Static, dynamic and fatigue behaviors of dental implant using finite element method. Advances in Engineering Software 37(10), 649–658. Kotoul, M., Skalka, P., Ševeček, O., Bertolla, L., Mertens, J., Marcián, P., hawla, N., 201 . rack bridging modelling in Bioglass ® based scaffolds reinforced by poly-vinyl alcohol/microfibrillated cellulose composite coating. Mechanics of Materials 110, 16–28. Marcián, P., Borák, L., Zikmund, T., Horáčková, L., Kaiser, J., Joukal, M., Wolff, J., 2021. On the limits of finite element models created from (micro)CT datasets and used in studies of bone-implant-related biomechanical problems. Journal of the Mechanical Behavior of Biomedical Materials 117, 104393. Marcián, P., Konečný, O., Borák, L., Valášek, J., Řehák, K., Krpalek, D., Florian, Z., 2011. On the Level of Computational Models in Biomechanics Depending on Gained Data from CT/MRI and MicroCT. MENDEL 2011, 255–267. Marcián, P., Lošák, P., Kaiser, J., Borák, L., 2016. Estimation of Orthotropic Mechanical Properties of Human Alveolar Bone. Engineering Mechanics 2016, 370–373. Misch C., 2008. Contemporary Implant Dentistry (3rd ed.). St Louis, United States: Elsevier - Health Sciences Division. Natali, A. N., Carniel, E. L., Pavan, P. G., 2008. Investigation of bone inelastic response in interaction phenomena with dental implants. Dental Materials 24(4), 561–569. Rahmoun, J., Auperrin, A., Delille, R., Naceur, H., Drazetic, P., 2014. Characterization and micromechanical modeling of the human cranial bone elastic properties. Mechanics Research Communications 60, 7–14. Schneider, C. A., Rasband, W. S., Eliceiri, K. W., 2012. NIH Image to ImageJ: 25 years of image analysis. Nature Methods 9(7), 671–675. Sekhon, K., Kazakia, G. J., Burghardt, A. J., Hermannsson, B., Majumdar, S., 2009. Accuracy of volumetric bone mineral density measurement in high-resolution peripheral quantitative computed tomography. Bone 45(3), 473–479. Shefelbine, S. J., Simon, U., laes, L., Gold, A., Gabet, Y., Bab, I., Müller, ., Augat, P., 2005. Prediction of fracture callus mechanical properties using micro-CT images and voxel-based finite element analysis. Bone, 36(3), 480–488. Ševeček, O., Bertolla, L., hlup, Z., Řehořek, L., Majer, Z., Marcián, P., Kotoul, M., 2019. Modelling of cracking of the ceramic foam specimen with a central notch under the tensile load. Theoretical and Applied Fracture Mechanics 100, 242–250. Ševeček, O., Papšík, ., Majer, Z., Kotoul, M., 2019. Influence of the cell geometry on the tensile strength of open-cell ceramic foams. Procedia Structural Integrity 23, 553–558. van Eijden TMGJ, van der Helm PN, van Ruijven LJ, Mulder L., 2006. Structural and Mechanical Properties of Mandibular Condylar Bone. Journal of Dental Research 85(1), 33-37. van Eijnatten, M., Koivisto, J., Karhu, K., Forouzanfar, T., Wolff, J., 2017. The impact of manual threshold selection in medical additive manufacturing. International Journal of Computer Assisted Radiology and Surgery 12(4), 607–615. van Ruijven, L. J., Giesen, E. B. W., Farella, M., van Eijden, T. M. G. J., 2003. Prediction of Mechanical Properties of the Cancellous Bone of the Mandibular Condyle. Journal of Dental Research 82(10), 819–823. Viceconti, M., 2011. Multiscale Modeling of the Skeletal System. Cambridge: Cambridge University Press. Apolena Šustková et al. / Procedia Structural Integrity 43 (2023) 276–281 281 Author name / Structural Integrity Procedia 00 (2022) 000–000 5 Figure 4: Values of Eapp in all directions (x,y,z – same as in Figure 2c)) for the homogeneous (on the left) and nonhomogeneous (on the right) model variants. Figure 5: Strain intensity in considered trabeculae (on the left) and along the path which was created based on the same start and end point (on the right). 4. Discussion Thresholding is a basic image processing tool for creating computational models from μ-CT images. The choice of the threshold fundamentally affects the results of stress-strain stages. As confirmed by the results of this study, the difference in the estimation of apparent mechanical properties is 74 % for homogeneous material and up to 59% for nonhomogeneous material when comparing the variants of models of geometry with minimum and maximum threshold values (0.25 and 0.65). In the case of a homogeneous material, the decrease in Eapp is linear and corresponds to the decrease in bone volume, which has the same value of E = 5000 MPa everywhere. When considering nonhomogeneous material, the decrease in Eapp with the decreasing volume for variants 0.25N, 0.30N and 0.35N is not so significant. This is due to the fact that the bone tissue at the edges does not have such a high density and its E, given by relation (1), is not so high, so there will be no significant decrease in Eapp as in the case of a homogeneous material. For other variants of nonhomogeneous material, the decrease in Eapp is more pronounced. Eapp values are relatively low for both variants of material, as it is a bone tissue sample with high porosity (83% to 90%). This corresponds to very low-density bone tissue (classification D4 according to Misch 2008). Nevertheless, values of Eapp obtained in this work correspond to the observations of the literature (van Eijden et al. 200 ; Marcián et al. 201 . From the comparison of the local stress of observed trabeculae, it is evident that the choice of the threshold has a more significant effect on the character of strain intensity distribution than the used material model. Character of the distribution for both selected material models is approximately the same. From the course of the strain intensity in the graph in Fig. 5 it can be seen that due to the change of the threshold value there is a substantial redistribution of strain. The difference of strain intensity between the models 0.25H and 0.65H is 55% and between the models 0.25N and 0.65N is 43%. 6 Author name / Structural Integrity Procedia 00 (2022) 000–000 5. Conclusion Using data from micro CT and image processing using automatic segmentation by thresholding, computational models were created and a comparative analysis was performed using micro FEM. The main observations of this study may be summarized as follows: 1. It is evident that the change of the threshold (the increasing value of threshold) has an impact on the amount of bone tissue. In this study, the difference in volumes between geometry models 0.25 and 0.65 is up to 38%. 2. The value of Eapp is decreasing with the increasing value of the threshold and thus with decreasing amount of the bone tissue for both material models. 3. Although the difference between the material used in this study is not so significant, the values of strain intensity and its distribution is dependent on the chosen threshold. Acknowledgement The work was supported by projects FSI-S-20-6164 and FSI-S-20-6175. References Borák, L., Marcián, P., 2017. Inhomogeneous Material Properties Assignment to Finite Element Models of Bone: A Sensitivity Study. Engineering Mechanics 2017, 190–193. osta, M. ., Tozzi, G., ristofolini, L., Danesi, V., Viceconti, M., Dall’Ara, E., 2017. Micro Finite Element models of the vertebral body: Validation of local displacement predictions. PLOS ONE 12(7), e0180151. Furukawa, T., Matsunaga, S., Morioka, T., Nakano, T., Abe, S., Yoshinari, M., Yajima, Y., 2019. Study on bone quality in the human mandibleAlignment of biological apatite crystallites. Journal of Biomedical Materials Research Part B: Applied Biomaterials 107(3), 838–846. Inagawa, H., Suzuki, N., Aoki, K., Wakabayashi, N., 201 . Potential for estimation of Young’s modulus based on computed tomography numbers in bone: A validation study using a nano-indentation test on murine maxilla. Dental, Oral and Craniofacial Research 4(4), 1-7. Kayabaşı, O., Yüzbasıoğlu, E., Erzincanlı, F., 2006. Static, dynamic and fatigue behaviors of dental implant using finite element method. Advances in Engineering Software 37(10), 649–658. Kotoul, M., Skalka, P., Ševeček, O., Bertolla, L., Mertens, J., Marcián, P., hawla, N., 201 . rack bridging modelling in Bioglass ® based scaffolds reinforced by poly-vinyl alcohol/microfibrillated cellulose composite coating. Mechanics of Materials 110, 16–28. Marcián, P., Borák, L., Zikmund, T., Horáčková, L., Kaiser, J., Joukal, M., Wolff, J., 2021. On the limits of finite element models created from (micro)CT datasets and used in studies of bone-implant-related biomechanical problems. Journal of the Mechanical Behavior of Biomedical Materials 117, 104393. Marcián, P., Konečný, O., Borák, L., Valášek, J., Řehák, K., Krpalek, D., Florian, Z., 2011. On the Level of Computational Models in Biomechanics Depending on Gained Data from CT/MRI and MicroCT. MENDEL 2011, 255–267. Marcián, P., Lošák, P., Kaiser, J., Borák, L., 2016. Estimation of Orthotropic Mechanical Properties of Human Alveolar Bone. Engineering Mechanics 2016, 370–373. Misch C., 2008. Contemporary Implant Dentistry (3rd ed.). St Louis, United States: Elsevier - Health Sciences Division. Natali, A. N., Carniel, E. L., Pavan, P. G., 2008. Investigation of bone inelastic response in interaction phenomena with dental implants. Dental Materials 24(4), 561–569. Rahmoun, J., Auperrin, A., Delille, R., Naceur, H., Drazetic, P., 2014. Characterization and micromechanical modeling of the human cranial bone elastic properties. Mechanics Research Communications 60, 7–14. Schneider, C. A., Rasband, W. S., Eliceiri, K. W., 2012. NIH Image to ImageJ: 25 years of image analysis. Nature Methods 9(7), 671–675. Sekhon, K., Kazakia, G. J., Burghardt, A. J., Hermannsson, B., Majumdar, S., 2009. Accuracy of volumetric bone mineral density measurement in high-resolution peripheral quantitative computed tomography. Bone 45(3), 473–479. Shefelbine, S. J., Simon, U., laes, L., Gold, A., Gabet, Y., Bab, I., Müller, ., Augat, P., 2005. Prediction of fracture callus mechanical properties using micro-CT images and voxel-based finite element analysis. Bone, 36(3), 480–488. Ševeček, O., Bertolla, L., hlup, Z., Řehořek, L., Majer, Z., Marcián, P., Kotoul, M., 2019. Modelling of cracking of the ceramic foam specimen with a central notch under the tensile load. Theoretical and Applied Fracture Mechanics 100, 242–250. Ševeček, O., Papšík, ., Majer, Z., Kotoul, M., 2019. Influence of the cell geometry on the tensile strength of open-cell ceramic foams. Procedia Structural Integrity 23, 553–558. van Eijden TMGJ, van der Helm PN, van Ruijven LJ, Mulder L., 2006. Structural and Mechanical Properties of Mandibular Condylar Bone. Journal of Dental Research 85(1), 33-37. van Eijnatten, M., Koivisto, J., Karhu, K., Forouzanfar, T., Wolff, J., 2017. The impact of manual threshold selection in medical additive manufacturing. International Journal of Computer Assisted Radiology and Surgery 12(4), 607–615. van Ruijven, L. J., Giesen, E. B. W., Farella, M., van Eijden, T. M. G. J., 2003. Prediction of Mechanical Properties of the Cancellous Bone of the Mandibular Condyle. Journal of Dental Research 82(10), 819–823. Viceconti, M., 2011. Multiscale Modeling of the Skeletal System. Cambridge: Cambridge University Press.