Gareth James, Contributing Writer11.17.15
In the medical device world—or in any new product development effort, for that matter—a sneak peek into the future would be tremendously helpful, as engineers could determine whether their device designs are optimal and encompass all variables. A trip to the future is unlikely, but the use of simulation in device design can help quell some of the research and development groups’ most nagging concerns.
Simulation offers many benefits when designing medical devices. For example, being able to accurately model designs under different conditions can allow researchers and engineers to better understand the mechanical behavior of devices, and also can lead to more tailored, patient-specific implants. When used as part of device prototyping, simulation technology can reduce time to market and generate valuable data about how an implant interacts with the body.
Regulatory concerns about simulation remain, however, while new technology solutions aim to reduce manual intervention and produce models and data suitable for long-term validation testing. While looking at current debates and the potential of a new hip implant-positioning tool, it is possible to review developments in medical device simulation.
Computational Modeling & Medical Devices
There are several key design and simulation approaches available to researchers, engineers, and analysts, including computer-aided design (CAD) and computer-aided engineering (CAE). CAE includes techniques such as finite element analysis (FEA)—a form of numerical modeling that calculates stresses and displacements under internal and external loads. In addition, computational fluid dynamics can analyze fluid flow patterns or be coupled to an FEA tool to predict fluid-structure interactions. Medical device manufacturers typically use these methods in early stage prototyping, enabling them to identify design deficiencies before they are locked in to the development process.
Acquiring 3-D imaging data from sources such as magnetic resonance imaging and computed tomography (CT) provides a basis for creating anatomical models of human anatomy. Image data can be used to reconstruct even the most complex anatomical structures of the body for use in computational simulation if taken at an appropriate resolution. To prepare image data for simulation, it is necessary to use image-processing software that can reconstruct the original geometry and segment regions of interest.
Image processing software such as ScanIP from Exeter, United Kingdom-based Simpleware Ltd. is used, in this respect, to generate 3-D models from scan data. With ScanIP, users are able to visualize and segment data, and obtain measurements and statistics. In terms of creating models suitable for orthopedic simulation, this software approach also includes capabilities for combining CAD objects, such as implants with image data, and exporting a wide range of multi-part meshes to finite element solvers.
Typical applications for this type of workflow include analyzing implant stability and tissue stress. The analyses can be carried out using either a limited number of bones or a large number of bones representative of the population. And in so doing, one can gain insights into specific physiologies and pathologies and how they may affect the performance of implants prior to manufacturing and clinical use. Quantitative information also can be collected, including threshold density for cortical and cancellous bone. This data can be used to quickly and efficiently develop devices, and has wide applications to parametric and patient-specific challenges when designing implants.
However, there are broader challenges associated with using simulation in medical device workflows. Hurdles can include accurately reproducing the complex structures of the human body for integration with devices; technical competence when designing and running simulations; consistency between results; and suitable validation against experimental data. In addition, the level of detail possible for a finite element model can be limited by the available computational resources, and by the difficulty of creating suitable models for population studies.
Regulatory Perspectives
While simulation is becoming more widely accepted by the U.S. Food and Drug Administration (FDA) as part of medical device submissions, it is still a contentious area in terms of the clarity in guidelines for use. The FDA’s Center for Devices and Radiological Health gradually has begun to focus on the potential for modeling to help in the review process. There has been hesitancy in using simulation in FDA submissions to validate medical device performance, as well in agreeing on common standards for modeling and collecting suitable data for in-silico clinical trials. Since 2011, the FDA has formed partnerships with software companies and other organizations to develop clearer standards for using computational simulation in device testing, and to collect more data and FDA-validated models usable as benchmarks.
Additionally, in cooperation with the FDA and industry, the American Society of Mechanical Engineers Verification and Validation in Computational Modeling of Medical Devices subcommittee, called V&V 40, is developing standards for verification and validation of simulations.
Early results from the FDA’s internal review of submissions that incorporate modeling by a team of internal modeling experts resulted in a draft guidance in 2014 on using simulation in medical device submissions. The FDA’s “Reporting Computational Modeling Studies in Medical Device Regulatory Submissions” summarizes reporting requirements for several physical modeling areas, including fluid dynamics and solid mechanics. For orthopedic researchers, guidance covers the need for clear descriptions of methodology, including the software employed, use of governing equations, physics simulated and boundary conditions used, as well as material properties and input from the literature.
The FDA’s guidelines remain a work in progress, though, with an ongoing need to improve communication between engineers, and to create usable data for verification and validation. Without consistencies between software approaches and the different workflows of research and development labs, coming up with a re-usable and reliable methodology will continue to be one of the major challenges for broadening the use of simulation. These types of problems are, however, being met by the work of organizations such as the Medical Device Innovation Consortium (MDIC).
The MDIC is a public-private partnership that was established in 2012 to address key regulatory challenges for the medical device industry. As part of the group’s effort to reduce barriers and increase clarity for the use of modeling in device submissions, the MDIC is looking to establish best practices for stakeholders involved in the manufacturing and approval of devices. Key goals include using simulation more widely to lower the cost of device trials, to collect data that might not be possible through bench testing, and to share data and resources that can be used by others for submissions.
New Automated Software Tools
A recent development in modeling medical devices and the human body has involved a collaboration between Simpleware, the University of Southampton, and ANSYS Inc. on a software methodology for predicting optimal positions for a total hip replacement. The project combined image processing, mesh generation, scripting, and simulation methods in order to demonstrate the potential to reduce development times when designing hip stems. The project aimed to understand the sensitivity of micromotion and other performance factors to hip implant positioning in the femur.
Ensuring that the stem is fitted in a pre-defined position to produce postoperative stability is crucial, but requires evaluating a large number of potential implants. For surgeons, this can be time-consuming, particularly when manual interventions are needed to achieve a perfect fit and to maximize bone-implant contact area. Poor bone quality also can cause cracking and the formation of gaps, leading to micromotion. Simulating the implant for a particular range of positions can reduce the risk of instability, as well as the impact of mismatches between the femoral canal and implant shapes, gaps at the interface, and intolerable bone-implant micromovements.
The tool developed by the project partners offers an alternative to time-consuming manual testing when re-positioning the implant and evaluating the effects of issues such as varus/valgus, ante/retroversion, anterior/posterior orientation, and implant offsets. Traditional CAD approaches typically require significant manual intervention to reposition the implant and replicate surgical procedures. Instead, using a finite element-based method and scripting, the newly developed tool can more efficiently account for minor adjustments and predict the optimum position for the implant. The addition of scripting and the generation of high-quality models integrating CAD and image data enables a robust, customizable method for exploring implant variability.
Optimizing Hip Implant Positioning
To test the tool, CT scan data of a 63-year-old male patient was acquired and segmented in Simpleware’s ScanIP software environment. The hip stem was virtually implanted in the medullary canal in a neutral (nominal) position using Simpleware software, with the implant based on the Exeter hip titanium alloy stem. The software also was used to generate tetrahedral meshes of the combined implant-femur model, taking into account bone density as being linearly related to the Hounsfield unit from the original CT scan—these computational models are suitable for analysis in finite element software such as ANSYS Mechanical APDL.
The Latin hypercube sampling option in ANSYS DesignXplorer was used to generate a design point table of 1,000 candidate positions, which was then reduced to 425 permissible implant positions. Material properties were assigned to each mesh element, and node sets created for simulating constraints and loading conditions from the femur and implant, with information taken from Simpleware software. Integrating CAD and image data was particularly useful, in this case, for removing the need to export and re-import anatomical and implant geometries, allowing multiple implant positions to be evaluated.
The meshes were used in ANSYS Mechanical APDL software to carry out micromotion and shear stress simulations under typical loading conditions such as walking and stair climbing. For the finite element analysis, the distal end of the femur was modeled as fixed, while joint and abductor loads were applied at a single node at the center of the top surface of the implant head and on the greater trochanter. Metrics used included average/maximum micromotion, percentage of implant area and bone interface (volume) average/maximum strain, and percentage of bone interface area with a strain larger than yield values.
Micromovements were calculated along the entire interface between the bone cavity and the external surface of the implant. A minimum von Mises stress of 144 megapascals was reported for the stem and the highest stresses found distally on the posterior side; the same trend was found for the micromotions, suggesting that when the femur is neutrally implanted, bone osseointegration will not be impeded, preventing formation of undesirable fibrous tissues.
The results from the simulations were collected into a response surface model using the Kriging regression method, with optimization algorithms used with the response surface to evaluate a full design space. This allowed for a prediction of optimal implant position, and an analysis of the relationship between micromotion and the sensitivity of individual placement variables. As a result, the tool demonstrates the significant promise of using an automated and integrated image processing, analysis, and model generation workflow for simulating device performance. The tool also offers potential for patient-specific modeling of implants, particularly when comparing different types of devices.
“We have already seen very positive results with the tool. We’re particularly pleased with how the scripting allows our users and software partners to cut down on the amount of manual time needed when trying to determine the best position for an implant,” said Philippe Young, Ph.D., managing director of Simpleware.
Marc Horner, Ph.D., technical leader for healthcare applications at ANSYS, noted that, “The scripting capability in Simpleware enables the ANSYS vision of robust design for implanted devices by automating the complex task of creating and then meshing assemblies of patient-specific anatomy and medical devices. The two-way communication between ANSYS Workbench and the Simpleware software suite allowed for the execution of a fully automated DOE study of the sensitivity of micromotion and other parameters to implant positioning. This represents one example of the limitless possibilities for medical device companies that require knowledge about the interaction between their devices and human anatomical structures.”
Future applications for the hip implant tool include assessing multiple implant designs for a specific femur, as well as considering factors in patient population studies, from implant width to size and other variables. Ongoing work between ANSYS, Simpleware, and the University of Southampton promises to develop the tool for new applications.
For example, recent research at the University of Southampton by Mamadou Bah, Ph.D., has explored how the tool can help understand the primary stability of cementless short stem implants across a range of patient morphologies, analyzing tolerance to subject variability.
Being able to use simulation to assist in pre-clinical planning, while still using experimental studies, aligns here with the broader trend of simulation as adding to the verification and validation tools available to surgeons. Although these software tools do not replace experimental testing, they provide a valuable and fast-evolving option for evaluating implant designs at an early stage of testing, allowing the effects of uncertainty and variability to be assessed.
As simulation becomes a more widely used option in orthopedics design, the ongoing challenge will be to exploit the speed and cost-savings of modeling alongside verification and validation tasks and long-term regulatory scrutiny. Finding ways to generate large amounts of data quickly and efficiently can arguably help with future testing and benchmarking of software for the FDA and other regulatory bodies.
Gareth James, Ph.D., works in public relations and marketing for Simpleware Ltd. in Exeter, U.K. He is involved in developing awareness of 3-D image-based modeling software and its benefits for orthopedic and other applications. He received his Ph.D. from the University of Exeter, and can be contacted at marketing@simpleware.com.
Simulation offers many benefits when designing medical devices. For example, being able to accurately model designs under different conditions can allow researchers and engineers to better understand the mechanical behavior of devices, and also can lead to more tailored, patient-specific implants. When used as part of device prototyping, simulation technology can reduce time to market and generate valuable data about how an implant interacts with the body.
Regulatory concerns about simulation remain, however, while new technology solutions aim to reduce manual intervention and produce models and data suitable for long-term validation testing. While looking at current debates and the potential of a new hip implant-positioning tool, it is possible to review developments in medical device simulation.
Computational Modeling & Medical Devices
There are several key design and simulation approaches available to researchers, engineers, and analysts, including computer-aided design (CAD) and computer-aided engineering (CAE). CAE includes techniques such as finite element analysis (FEA)—a form of numerical modeling that calculates stresses and displacements under internal and external loads. In addition, computational fluid dynamics can analyze fluid flow patterns or be coupled to an FEA tool to predict fluid-structure interactions. Medical device manufacturers typically use these methods in early stage prototyping, enabling them to identify design deficiencies before they are locked in to the development process.
Acquiring 3-D imaging data from sources such as magnetic resonance imaging and computed tomography (CT) provides a basis for creating anatomical models of human anatomy. Image data can be used to reconstruct even the most complex anatomical structures of the body for use in computational simulation if taken at an appropriate resolution. To prepare image data for simulation, it is necessary to use image-processing software that can reconstruct the original geometry and segment regions of interest.
Image processing software such as ScanIP from Exeter, United Kingdom-based Simpleware Ltd. is used, in this respect, to generate 3-D models from scan data. With ScanIP, users are able to visualize and segment data, and obtain measurements and statistics. In terms of creating models suitable for orthopedic simulation, this software approach also includes capabilities for combining CAD objects, such as implants with image data, and exporting a wide range of multi-part meshes to finite element solvers.
Typical applications for this type of workflow include analyzing implant stability and tissue stress. The analyses can be carried out using either a limited number of bones or a large number of bones representative of the population. And in so doing, one can gain insights into specific physiologies and pathologies and how they may affect the performance of implants prior to manufacturing and clinical use. Quantitative information also can be collected, including threshold density for cortical and cancellous bone. This data can be used to quickly and efficiently develop devices, and has wide applications to parametric and patient-specific challenges when designing implants.
However, there are broader challenges associated with using simulation in medical device workflows. Hurdles can include accurately reproducing the complex structures of the human body for integration with devices; technical competence when designing and running simulations; consistency between results; and suitable validation against experimental data. In addition, the level of detail possible for a finite element model can be limited by the available computational resources, and by the difficulty of creating suitable models for population studies.
Regulatory Perspectives
While simulation is becoming more widely accepted by the U.S. Food and Drug Administration (FDA) as part of medical device submissions, it is still a contentious area in terms of the clarity in guidelines for use. The FDA’s Center for Devices and Radiological Health gradually has begun to focus on the potential for modeling to help in the review process. There has been hesitancy in using simulation in FDA submissions to validate medical device performance, as well in agreeing on common standards for modeling and collecting suitable data for in-silico clinical trials. Since 2011, the FDA has formed partnerships with software companies and other organizations to develop clearer standards for using computational simulation in device testing, and to collect more data and FDA-validated models usable as benchmarks.
Additionally, in cooperation with the FDA and industry, the American Society of Mechanical Engineers Verification and Validation in Computational Modeling of Medical Devices subcommittee, called V&V 40, is developing standards for verification and validation of simulations.
Early results from the FDA’s internal review of submissions that incorporate modeling by a team of internal modeling experts resulted in a draft guidance in 2014 on using simulation in medical device submissions. The FDA’s “Reporting Computational Modeling Studies in Medical Device Regulatory Submissions” summarizes reporting requirements for several physical modeling areas, including fluid dynamics and solid mechanics. For orthopedic researchers, guidance covers the need for clear descriptions of methodology, including the software employed, use of governing equations, physics simulated and boundary conditions used, as well as material properties and input from the literature.
The FDA’s guidelines remain a work in progress, though, with an ongoing need to improve communication between engineers, and to create usable data for verification and validation. Without consistencies between software approaches and the different workflows of research and development labs, coming up with a re-usable and reliable methodology will continue to be one of the major challenges for broadening the use of simulation. These types of problems are, however, being met by the work of organizations such as the Medical Device Innovation Consortium (MDIC).
The MDIC is a public-private partnership that was established in 2012 to address key regulatory challenges for the medical device industry. As part of the group’s effort to reduce barriers and increase clarity for the use of modeling in device submissions, the MDIC is looking to establish best practices for stakeholders involved in the manufacturing and approval of devices. Key goals include using simulation more widely to lower the cost of device trials, to collect data that might not be possible through bench testing, and to share data and resources that can be used by others for submissions.
New Automated Software Tools
A recent development in modeling medical devices and the human body has involved a collaboration between Simpleware, the University of Southampton, and ANSYS Inc. on a software methodology for predicting optimal positions for a total hip replacement. The project combined image processing, mesh generation, scripting, and simulation methods in order to demonstrate the potential to reduce development times when designing hip stems. The project aimed to understand the sensitivity of micromotion and other performance factors to hip implant positioning in the femur.
Ensuring that the stem is fitted in a pre-defined position to produce postoperative stability is crucial, but requires evaluating a large number of potential implants. For surgeons, this can be time-consuming, particularly when manual interventions are needed to achieve a perfect fit and to maximize bone-implant contact area. Poor bone quality also can cause cracking and the formation of gaps, leading to micromotion. Simulating the implant for a particular range of positions can reduce the risk of instability, as well as the impact of mismatches between the femoral canal and implant shapes, gaps at the interface, and intolerable bone-implant micromovements.
The tool developed by the project partners offers an alternative to time-consuming manual testing when re-positioning the implant and evaluating the effects of issues such as varus/valgus, ante/retroversion, anterior/posterior orientation, and implant offsets. Traditional CAD approaches typically require significant manual intervention to reposition the implant and replicate surgical procedures. Instead, using a finite element-based method and scripting, the newly developed tool can more efficiently account for minor adjustments and predict the optimum position for the implant. The addition of scripting and the generation of high-quality models integrating CAD and image data enables a robust, customizable method for exploring implant variability.
Optimizing Hip Implant Positioning
To test the tool, CT scan data of a 63-year-old male patient was acquired and segmented in Simpleware’s ScanIP software environment. The hip stem was virtually implanted in the medullary canal in a neutral (nominal) position using Simpleware software, with the implant based on the Exeter hip titanium alloy stem. The software also was used to generate tetrahedral meshes of the combined implant-femur model, taking into account bone density as being linearly related to the Hounsfield unit from the original CT scan—these computational models are suitable for analysis in finite element software such as ANSYS Mechanical APDL.
The Latin hypercube sampling option in ANSYS DesignXplorer was used to generate a design point table of 1,000 candidate positions, which was then reduced to 425 permissible implant positions. Material properties were assigned to each mesh element, and node sets created for simulating constraints and loading conditions from the femur and implant, with information taken from Simpleware software. Integrating CAD and image data was particularly useful, in this case, for removing the need to export and re-import anatomical and implant geometries, allowing multiple implant positions to be evaluated.
The meshes were used in ANSYS Mechanical APDL software to carry out micromotion and shear stress simulations under typical loading conditions such as walking and stair climbing. For the finite element analysis, the distal end of the femur was modeled as fixed, while joint and abductor loads were applied at a single node at the center of the top surface of the implant head and on the greater trochanter. Metrics used included average/maximum micromotion, percentage of implant area and bone interface (volume) average/maximum strain, and percentage of bone interface area with a strain larger than yield values.
Micromovements were calculated along the entire interface between the bone cavity and the external surface of the implant. A minimum von Mises stress of 144 megapascals was reported for the stem and the highest stresses found distally on the posterior side; the same trend was found for the micromotions, suggesting that when the femur is neutrally implanted, bone osseointegration will not be impeded, preventing formation of undesirable fibrous tissues.
The results from the simulations were collected into a response surface model using the Kriging regression method, with optimization algorithms used with the response surface to evaluate a full design space. This allowed for a prediction of optimal implant position, and an analysis of the relationship between micromotion and the sensitivity of individual placement variables. As a result, the tool demonstrates the significant promise of using an automated and integrated image processing, analysis, and model generation workflow for simulating device performance. The tool also offers potential for patient-specific modeling of implants, particularly when comparing different types of devices.
“We have already seen very positive results with the tool. We’re particularly pleased with how the scripting allows our users and software partners to cut down on the amount of manual time needed when trying to determine the best position for an implant,” said Philippe Young, Ph.D., managing director of Simpleware.
Marc Horner, Ph.D., technical leader for healthcare applications at ANSYS, noted that, “The scripting capability in Simpleware enables the ANSYS vision of robust design for implanted devices by automating the complex task of creating and then meshing assemblies of patient-specific anatomy and medical devices. The two-way communication between ANSYS Workbench and the Simpleware software suite allowed for the execution of a fully automated DOE study of the sensitivity of micromotion and other parameters to implant positioning. This represents one example of the limitless possibilities for medical device companies that require knowledge about the interaction between their devices and human anatomical structures.”
Future applications for the hip implant tool include assessing multiple implant designs for a specific femur, as well as considering factors in patient population studies, from implant width to size and other variables. Ongoing work between ANSYS, Simpleware, and the University of Southampton promises to develop the tool for new applications.
For example, recent research at the University of Southampton by Mamadou Bah, Ph.D., has explored how the tool can help understand the primary stability of cementless short stem implants across a range of patient morphologies, analyzing tolerance to subject variability.
Being able to use simulation to assist in pre-clinical planning, while still using experimental studies, aligns here with the broader trend of simulation as adding to the verification and validation tools available to surgeons. Although these software tools do not replace experimental testing, they provide a valuable and fast-evolving option for evaluating implant designs at an early stage of testing, allowing the effects of uncertainty and variability to be assessed.
As simulation becomes a more widely used option in orthopedics design, the ongoing challenge will be to exploit the speed and cost-savings of modeling alongside verification and validation tasks and long-term regulatory scrutiny. Finding ways to generate large amounts of data quickly and efficiently can arguably help with future testing and benchmarking of software for the FDA and other regulatory bodies.
Gareth James, Ph.D., works in public relations and marketing for Simpleware Ltd. in Exeter, U.K. He is involved in developing awareness of 3-D image-based modeling software and its benefits for orthopedic and other applications. He received his Ph.D. from the University of Exeter, and can be contacted at marketing@simpleware.com.