Mechanical interaction between human soft tissue and elastic supports

Thread 1. Computational Methods in Biomechanics and Mechanobiology 6824 Th, 08:30-08:45 (P39) Geometric intra-subject variability of arm vessels assessed by MRA: A challenge for quantification and modeling of the vascular access for hemodialysis L. Antiga 1, N. Planken 2, M. Piccinelli 1, B. Ene-lordache 1, W. Huberts 3, A. Remuzzi 1, J. Tordoir2 . 1Bioengineering Department, Marie Negri Institute,

Bergamo, Italy, 2Department of Surgery, Maastricht University Hospital, Maastricht, The Netherlands, 3Department of Biomedical Engineering, Eindhoven University of Technology, Eindhoven, The Netherlands Predicting maturation and long-term patency of a vascular access for hemodialysis has become a crucial issue for patient treatment and morbidity. Measurements of vessel caliber have been employed in the attempt of predicting maturation, and hemodynamic studies have been conducted to identify regions of disturbed flow potentially associated with intimal hyperplasia. However, the inherent variability of the arm vasculature potentially poses a threat to the use of absolute measurements of vessel size or local hemodynamic parameters for clinical predictions. In this work, we investigate intra-subject variability of the vasculature of the left arm in one healthy volunteer (A) and in two patients with a native end-to-side radio-cephalic (B) and brachio-basilic (C) fistula, respectively, using serial CEMIRA acquisitions. Three CE-MRA scans where performed on subject A, the first two under the same physiologic conditions, the third after a protein-rich meal. Subject B underwent two scans, before and after a hemodyalitic session, respectively. Subject C underwent two scans several months apart. 3D models of the arm vasculature were reconstructed with a level set approach, using a novel gradient-based advection field suited for the segmentation of small vessels. Centerlines were computed using a Voronoi-based minimal cost path approach, and models were parameterized by centerline abscissas, generating correspondence maps between vessel surfaces from successive scans. This way, variability in local vessel radius could be quantified continuously over entire vascular trees. The results shows that the vasculature of the arm is subject to changes in size up to 20% in all subjects, and that different branches in the same vascular tree can experience different caliber changes, even under similar physiologic conditions. This potentially poses a challenge to absolute quantifications of size-related quantities below a given precision, but, at the same time, it opens new perspectives for the assessment of vascular adaptivity by means of distributed differential evaluations. 7495 Th, 08:45-09:00 (P39) Implementation of non-linear, large deformation, dynamic models for traumatic events simulation on thorax and spine C. Frigo, E.E. Pavan, M. Achilli, S. Brugnettini. TBM Lab, Laboratory ef

Movement Biomechanics and Motor Control, Department of Bioengineering, Polytechnic of Milan, Milan, Italy This work was aimed at developing a Finite Element (FE) model of head and trunk to simulate the effects of large accelerations and impacts. The intended application was in the field of forensic biomechanics and a preliminary implementation was designed with reference to car accidents. The FE model was built by a segmentation of human body slices obtained from the Visible Human Dataset (National Library of Medicine). The morphology of human organs was obtained by means of a segmentation software (Amira). More than one hundred different components of the human body were reconstructed in 3-D including bones, cartilage structures of thorax and cervical spine and the vital organs inside the rib cage. The external shape of each anatomical component was modelled by an FE software tool and the mechanical and structural properties of the anatomical component (density, viscous-elastic properties) were defined on the basis of data available in literature. The whole model contains more than one million of elements and its simulation appeared very heavy from the computational point of view. For this reason the model was split into different functional units, and different loading conditions were analysed separately. In the present work the attention was focused on the effect of whiplash on the cervical spine and of external objects impacting the thorax. The model was implemented with Hyper-Mesh software as a preprocessor, Ls-Dyna as a solver, and Hyper-View as post-processor. Concerning whiplash the cervical/thoracic spine was rigidly connected to a base to which an acceleration law was imposed. In two different simulations the constrained vertebrae were respectively T1 and T4. The relevant output was the flexion-extension angle of the vertebral units and the load distribution in the intervertebral disks and cartilage. The last test concerned a frontal impact on the rib cage and the objective was to estimate the level of deformation of the cage and the consequent risk of fracture and damage for internal organs. The obtained results were critically analysed in relation to previous data available in literature.

T1.14 Image-based Anatomical Modelling for CAD/FEA Applications

$429

7701 Th, 09:00-09:15 (P39) Development o f 3-D finite element model of lumbosacral vertebral column A.B. Deoghare, P.M. Padole. Department ef Mechanical Engineering,

Visvesvarya National Institute of Tchnology, Nagpur Maharashtra, India Finite element analysis is becoming an increasingly important part of biomechanics and Orthopedic research, as computational resources become more powerful and data handling algorithms become more sophisticated. Until recently, tools with sufficient power did not exist or were not accessible to adequately model complicated, three-dimensional, nonlinear biomechanical systems. In the past, finite element analyses in biomechanics have often been limited to two-dimensional approaches, linear analyses, or simulations of single tissue types. The resources to model fully three-dimensional, nonlinear, multitissue, and even multi-joint systems. The authors will present the process of developing these kinds of finite element models, considering the example of human lumbosacral vertebra. 6122 Th, 09:15-09:30 (P39) Analysis o f a probabilistic shape-based femur model T.L. Bredbenner 1, K.A. Bartels 2, L.M. Havill 3, D.P. Nicolella 1. 1Materials

Engineering Department, Southwest Research Institute, San Antonio, TX, USA, 2Medical Systems Department, Southwest Research Institute, San Antonio, TX, USA, 3Department of Genetics, Southwest Foundation for Biomedical Research, San Antonio, TX, USA The development of finite element (FE) models that accurately describe the complex morphology of biological structures is challenging; model development is typically much more costly than model solution. To address this problem, we implemented statistical shape modeling methodology to rapidly construct high-fidelity FE models of the complex geometry and material property distribution of skeletal structures from imaging data. Surface geometries were generated from computed tomography data for a training set of femurs, and a computational FE mesh was projected on to each surface to create volumetric mesh geometries (TrueGrid, XYZ Scientific Corp.). Apparent density was determined from imaging data at the spatial location of each mesh node (MATLAB, The Mathworks). Correlation between spatial positions and density values for each mesh node was described using a covariance matrix. Point to point correspondence between individual meshes was achieved by optimizing the location of mesh points to minimize the sum of the squares of the eigenvalues of the covariance matrix. A Principal Components Analysis (PCA) decomposition of the covariance matrix allowed description of each femur based on the mean geometry and bone density distribution and the PCAdetermined eigenvalues and eigenvectors. PCA demonstrated that 99.94% of the variation in the geometry and bone density distribution of the femur set was described by a single eigenvalue. Probabilistic analysis of the effects of femur geometry, bone density variation, and the error associated with determining material properties from bone density demonstrated the effect of model uncertainty on FE model predictions. The probabilistic analysis demonstrates the variation in stress response predictions that occur as a result of variation in femur geometry and bone material property distribution. Furthermore, the variability in shape and bone density distribution is represented by a minimal number of uncorrelated random variables, reducing the computational burden of performing a probabilistic analysis. 5800 Th, 09:30-09:45 (P39) Mechanical interaction between human soft tissue and elastic supports G. Silber 1, M. Schrodt 1, G. Benderoth 1, C. Then 1, J.O. Balzer 2, T.J. Vogl 2.

Center of Biomedical Engineering, Frankfurt~Main, Germany; 1Fachhochschule Frankfurt, Institut fEtr Materialwissenschaften, Frankfurt~Main, Germanf 2Diagnostische und Interventionelle Radiologie, Klinikum der Johann Wolfgang Goethe-Universit&t Frankfurt, Frankfurt~Main, Germany The gap between rising demands for health care caused by the demographic development of western countries and political claims for reducing the costs of health care systems, can be bridged among other things by optimizing care taking procedures. Avoiding bed sore (decubitus ulcer) is a current and will become a rising problem because of the increasing lifespan of the people of our societies. If a method for designing supports is developed, to extent the time to move a patient by the nursing staff for avoiding decubitus, it will be a step in this optimization procedure. Besides the genetic and physiological aspects the stress distribution in the sub dermal tissue plays a decisive role by the occurrence of decubitus ulcer. A method to determine this stress distribution and some stress and strain calculations with different elastic supports is given in this contribution. The method consists of a loading device able to load various parts of a living human body in a reproducible way. The loading test were carried out in a MRI to be able to reconstruct the complex geometry of the tissue in an unloaded and loaded state of the chosen test site (the heel in this study). By using

$430

Journal of Biomechanics 2006, Vol. 39 (Suppl 1)

Ogden-Hill-type constitutive equations, a multi objective quality function and an FE-tool, parameters were determined for the different tissues. For calculating the stress distribution at the test site an FE-model was developed. It was built up by elastic supports consisting of various soft foams with different geometries and a norm body including the previously determined test site. Calculations were made for different material and geometry combinations of the support. The model will be extended with modified slatted frames (micro stimulation) to verify possible benefits of such systems.

T1.15 Biological Flows T1.15.1 Particle Tracking and Particle Methods for Biological Flows 5838 Mo, 08:15-08:30 (PT) Modelling particle deposition in a turbulent airway model M.A.I. Khan 1, X.'~ Luo 1, F.C.G.A. Nicolleau2. 1Department of Mathematics, University of Glasgow, Glasgow, UK, 2Department of Mechanical Engineering, University of Sheffield, Sheffield, UK Transport and deposition of aerosol particles in a stenotic tube model are studied to investigate the effects of turbulent flow structure on particle deposition. We introduce a novel particle tracking model namely kinematic simulation (KS) [1], which is a Lagrangian model of turbulent dispersion that takes into account the effects of spatio-temporal flow structure on particle dispersion. It is a unified Lagrangian model of one-, two- and indeed multi-particle turbulent dispersion and can easily be used as a Lagrangian sub-grid model for large eddy simulation (LES) code thus enabling complex geometry to be taken into account. To study the effect of small scale flow structure on particle deposition in the stenotic pipe flow we use LES to simulate the flow filed, and KS to model the sub-grid flow structure to generate particle trajectories. Thus the large scales are resolved by the simulation and the small scales are modelled using various sub-grid models. As none of the existing sub-grid models are known to have taken into account the effects of small-scale turbulent flow structures on particle deposition, it is important to use KS's ability to re-model the subgrid velocity field and thereby incorporate its effect on particle deposition. The parameters of our simulations for LES are the Reynolds number, diameter of the pipe, percentage of stenoses and sub-grid model parameters. For KS the parameters are the energy dissipation rate obtained from LES, the energy spectra, ratio of the largest and smallest sub-grid scales and the total number of modes for the sub-grid velocity field. The turbulent flow features thus obtained are compared with published experimental data [2] in a stenotic pipe. Preliminary results suggest that the particle deposition in the stenotic tube can be greatly affected by the small-scale (sub-grid) turbulent flow structures. References [1] JCH Fung et al. J Fluid Mech 1992; 236: 281. [2] SA Ahmed et al. J Biomech 1983; 16: 505. 6497 Mo, 08:30-08:45 (PT) Lattice-Boltzmann calculations of blood flow in a fluidised bed: results for the permeability and drag A.K.M. Podias, Y.F. Missirlis. Biomedical Engineering Laboratory, Mechanical Engineering and Aeronautics Department, University of Patras, Rion-Patras, Greece The flow and transport within fluidised beds depends strongly on particleparticle and fluid-particle interaction. This is the reason that proper closure relations for these two interactions are vital for reliable predictions on the basis of continuum models. In a previous study [Podias A.K.M. and Missirlis '~E (2000), In: Prendergast EJ., Lee T.C., Carr A..J. (Eds.), Proceedings of the 12th Conference of the European Society of Biomechanics, Royal Academy of Medicine in Ireland], a model for describing the blood flow field and heparin transport within a heparin-adsorbing device operating as a fluidised bed has been derived and calculated using the Finite Element Method. There, the fluid dynamic interactions between particles in the multi-particle assemblage have been accounted for by employing the so-called sphere-in-cell model [Happel J. (1958), AIChE J. 4, 197]. The present study, demonstrates the use of the lattice-Boltzmann equation (LBE) method in predicting low Reynolds number flow past a mono-dispersed and random assemblage of rigid, fluidised, spherical particles, thereby focusing in the blood-particle interaction relation. The LBE method employed is fully explicit and time-dependent, in which distribution of fluid particles exists at discrete locations in space and move in discrete directions, speeds and intervals of time. Flow quantities such as density and velocity are defined as the appropriate moments over the state space of the distribution values at a given node and time step. The particle distribution dynamics via the application of discrete kinetic theory provides full recovery of the Navier-Stokes continuum fluid equations for the behaviour of the macroscopic fluid properties.

Oral Presentations Estimates of the local velocity distributions and permeability are obtained for wide ranges of physical and kinematic conditions. Results on the dynamics of the flow in terms of the resulting drag coefficient are also obtained and discussed. The predicted permeability and drag coefficient is compared with theoretical estimates from the literature and with our experimental results [Podias A.K.M. and Missirlis '~E (2002), In: R. Bedzinski, C. Pezowicz, K. Scigala (Eds.), ACTA of Bioengineering and Biomechanics: Proceedings of the 13th Conference of the European Society of Biomechanics, Wroclaw, Poland]. 5071 Mo, 08:45-09:00 (P7) Particle simulations of blood flow in vein with many red blood cells K. Nagayana. Department of Mechanical Information Science & Technology Kyushu Institute of Technology, lizuka, Japan Particle simulations of blood flow in vein with red blood cells were carried out. This model considers plasma as fluid particles with viscous forces and red blood cell as elastic particles using springs for surface and fluid particles inside. Vein is modeled as solid particles. 2D and 3D simulations are carried out. In 2D flow between parallel walls, more than 30 RBCs are simulated for cases with and without narrow position assuming thrombus. Between parallel walls, RBCs tend to turn parallel to the flow direction and flows away from the wall to reduce flow resistance. With narrow position, interactions among red blood cells increased. At very low Re number region, recirculation was not appear, and RBC was not captured around narrow position. The model also extended to simplified case with thrombus formation, and the preliminary results were obtained. As thrombus grew, flow resistance increased and blood flow rate decreased. Three dimensional particle simulations of blood flow with red blood cell are also carried out. First plasma flow was tested comparing with theoretical Pouiseille flow. Red blood cell shape was modeled as sphere at first, and removing plasma particles inside, the shape change is checked. And finally, blood flow with blood cell inside the vein was simulated for cases from one RBC to several RBCs. In case of a RBC flowing at center of the vein, parachute type deformation was observed. In case of several RBCs, their interaction was studied. 6576 Mo, 09:00-09:15 (P7) Three-dimensional simulations of microscopic blood flow using SPH method N. Tanaka, Y. Hayakawa, T. Masuzawa. Department ef Mechanical Engineering, Ibaraki University, Hitachi, Japan We have developed microscopic blood model based on the smoothed particle hydrodynamics (SPH) method. In the model, plasma fluid is discretized by SPH particles, and a red blood cell (RBC) is expressed by internal SPH particles surrounded by elastic membrane (structure) particles. In addition, a new interaction model between fluid particles and structure particles has been developed in order to prevent an internal particle from getting out of the membrane. This model is also applicable to the interaction between fluid particles and wall particles for preventing a fluid particle from moving into wall. For verifying the model, we numerically analyzed the three-dimensional tanktread motion of an RBC under a constant shear field. The numerical results can well reproduce behaviors of RBC. We also analyzed another numerical example of blood flow in stenosed vessel. The results show that a RBC flexibly changes its shape according to the vessel geometry and moves past the narrow vessel part. Finally, the numerical results were visualized by the recent ray-tracing technique for the purpose of realistic representation. 6790 Mo, 09:15-09:30 (P7) Simulation study on effects of elastic red blood cells on primary thrombogenesis using particle method K.-i. Tsubota, H. Kamada, S. Wada, T. Yamaguchi. Department ef Bioengineering and Robotics, Graduate School of Engineering, Tohoku University, Sendal, Japan A novel computer simulation approach using particle method was proposed to analyze the formation of primary thrombus due to platelet aggregation in the blood flow. Platelets, elastic red blood cells (RBCs) and plasma fluid, which are main components of blood, were modeled by discrete particles. The platelet aggregation to the injured vessel wall was expressed by introducing an attractive force transferred from the injured wall to the platelets. The solidlike mechanical properties of the primary thrombus consisting of aggregated platelets were expressed by spring force acting between the adhered platelet particles. The particles for the RBC membrane were connected with their neighboring membrane particles by stretch/compression and bending springs. Being subjected to an incompressible viscous flow governed by Navier-Stokes (N-S) equations, the motion of all the particles was solved by using the MPS method. The forces induced by the springs that act on the particles for the