Medical Device Link.

Originally Published MDDI January 2004

Software Modeling

Figure 1. Models generated with MSC.Patran of a 2-D cross section of artery, stent, and angioplasty balloon. The image on the left is before expansion of the angioplasty balloon and the image on right is after the expansion (Click to Enlarge).

Contrary to historical schools of thought, all heart attacks are not caused by the gradual narrowing of one or more coronary artery segments or the slow progressive development of flow-limiting coronary lesions. An unstable lesion, which has been classified as a fibroatheroma, can rupture due to unexpected triggers and, according to recent research, can also cause heart attacks.

Testing interventional therapies for treating a fibroatheroma can be expensive and time-consuming. Creating a benchtop model of arterial soft tissue, which has multiple constituents, presents a challenge. However, finite-element (FE) modeling technologies enable computer modeling of soft tissues and interventional devices. These models allow medical device design engineers to evaluate new concepts in treating a human fibroatheroma.

It can be difficult to imagine how to construct a benchtop model that allows engineers to evaluate whether a device being developed is having a positive or a negative impact on a fibroatheroma. FE modeling of the soft tissue constituents characteristic of a fibroatheroma provides a tool that facilitates the development of such a benchtop model. It also provides a mechanism by which engineers can evaluate the performance of any device or concept (see Figure 1).

Computational modeling of arterial soft tissue is an inherently complex task that can be simplified by making educated assumptions about the material, the geometry, or both. For example, the complex nonlinear behavior of the arterial tissue material may be represented as a linear elastic material based on assumptions about the physiological loading range of interest. However, educated assumptions can affect the breadth to which the results apply. Ensuring the accuracy of results over a larger dynamic range of responses may require addressing other factors, including:

• More-complex soft tissue behavior, such as nonlinear elasticity or hyperelasticity.
• Tissue plasticity or viscoelastic effects, such as creep (sustained deformation under constant load).
• Stress relaxation (diminishing load under a constant deformation state). 

Adding an interventional device to the analysis increases the complexity of the model. The intricate behavior of the device and the complications associated with the introduction of contact between the tissue and the device add layers of complexity. Despite the application of any simplifying assumptions about the response of the tissue or the behavior of the device, analyzing the interaction between the two is complex and thus requires nonlinear analysis.

Geometry Creation

Creating geometry for soft tissue models requires more than importing the geometry of a well-defined mechanical part from computer-aided design (CAD) software. Typically soft tissue and arterial geometry evolve from images taken with one of many medical imaging systems. Guidant’s New Ventures Group employs typical methodologies for extracting data on the geometric structure and morphology of soft tissue. These methodologies are based on medical imaging modalities such as intravascular ultrasound (IVUS), optical coherence tomography (OCT), or magnetic resonance imaging (MRI). These systems yield two-dimensional images, three-dimensional data sets, or a series of 2-D images taken along a particular axis, which can be reconstructed into 3-D data sets (see Figure 2). 

The data generated by any of these imaging modalities typically take the form of a bitmap image and can be processed in one of many paint or image processing applications. Most commercial image processing packages, such as Adobe Photoshop, have been developed for the art and graphics industry. However, the image analysis capabilities in these programs have an unexpected application in the bioengineering arena. These commercial packages can provide the foundation for creating geometric models of soft tissue based on the data sets exported from a medical imaging system. These software applications allow isolation of features, and they can highlight particular constituents within the wall of the coronary artery. These constituents play an important role in the final geometric model.

If soft tissue models are destined for engineering analysis, the bitmap images must be converted into a vector-based geometry through a process that is generally referred to as tracing. Tracing coaxes data into a form that the FE modeler can recognize. A host of applications and techniques from both the academic and commercial communities exist to facilitate this operation, and each accomplishes this task in a different manner.

The automatic tracing capabilities may be limited in some commercially available graphics applications. These commercial programs can provide some simple tools for handling this conversion by allowing a user to trace highlighted features within a bitmap image. 

When working with a few images in a 2-D space, a manual tracing process is not very cumbersome; however, as the data sets and analyses are extended to 3-D or the number of 2-D images increases, automated processes must be explored. Looking at some of the software employed to accomplish these tasks, it is easy to see that some aspects of soft tissue mechanics are a blend of the artistic and engineering worlds. The process begins with medical images of the free-form structures characteristic of soft tissues and transforms them into geometric objects for engineering analysis.

With the conversion of tissue components of interest into a vector-based data set or into a solid model, the geometry can be imported into an FE modeler. 

Finite-Element Analysis

In FE analysis, the models are divided into pieces to address various objectives of the simulation. Once the geometry is opened in the FE modeler, the basic arterial geometry can be broken into pieces to apply specific boundary conditions or to define specific interactions. Essentially, one proceeds with the standard model construction process as if building a model of any other mechanical device.

In the world of soft tissue mechanics, however, there are many more free-form shapes, which present a challenge when working with traditional CAD or modeling software packages. Therefore, a significant amount of preparatory work is required prior to importing geometry into the FE modeler.

With a soft tissue model in place, the focus of model development shifts to the device with which the tissue is to interact. Ultimately, the objectives of the finite-element analysis drive the origin of device geometry. The geometry can originate from a CAD model or it can be extracted from a previous analysis. It can even be created on the fly in a modeler such as MSC.Patran to perform a preliminary evaluation of a new concept or interaction.
In the case of modeling the interaction between arterial soft tissue and an interventional device, the analysis may demand device geometry as simple as a cross section of an angioplasty balloon or a stent to perform a particular 2-D analysis. 

Regardless of its source, a certain amount of processing is typically required to extract the specific device geometry pertinent to the analysis. At times, generating the appropriate geometry data can become an arduous and time-consuming task.

Figure 2. Image from intravascular ultrasound (IVUS), one of several medical imaging modalities used to begin the process of developing a soft tissue engineering model (Click to Enlarge).

An embedded capability in some modelers allows a user to create scripts to automate a variety of operations. In the Guidant example, a series of proprietary Patran command language (PCL) routines have been written to generate stent geometries for use in a variety of analyses. The automation provided by these custom PCL scripts enabled Guidant to gain a clear understanding of how a device will affect the tissue. The scripts allowed the company to efficiently evaluate a number of different stent parameters.

To create a matrix of information, many different stent parameters can be modeled to evaluate how the stents are affecting the tissue. FE modelers provide the means to write scripts that execute some processes automatically, such as performing complex analyses or creating device models with every design perturbation. The scripts eliminate the need to manually step through the analytical processes, which can be time-consuming.

Material Properties of Soft Tissue

Many different material models of soft tissues have been developed. These models include simple linear elastic materials, hyperelastic materials, viscoelastic materials, and even poroelastic materials. 

As with most soft tissue analyses, compromises must be made based on an evaluation of costs versus benefits. For example, a linear elastic representation of arterial tissue is both simple and computationally less expensive than a hyperelastic model.

Fundamentally, a linear elastic material requires a single number for a complete material definition. However, this representation may inadequately represent the response of an artery over the entire range of deformations that occur during delivery of an interventional device.

A hyperelastic model or a material model with plasticity may provide a better representation of arterial behavior over the entire range of deformation to which an artery might be subjected during device delivery. However, these material structures are computationally more expensive. In addition, increasing the complexity of the material model also adds to the costs associated with developing and running more-complex experiments to define multiple material parameters or coefficients.

Loading

Defining loads can be an overwhelmingly complex task if all of the loads that affect the arterial tissue and an interventional device, such as a stent, are taken into account. Typically, an analysis can be simplified by strategically selecting the loads that dominate the response. 

If intuition and experience do not provide enough justification for including or excluding a particular load, small parametric studies can often help determine which loads are important to include in the final analysis. Nonetheless, the simplest starting point is to look at the physical problem being modeled and identify the important loading conditions for each component.

Focusing on the delivery of a stent, three primary components—an angioplasty balloon, a stent, and the coronary artery—are interacting at the site of a coronary lesion when a stent is deployed. Reviewing the dynamics of an interventional procedure suggests three dominant loads need to be considered. Those dominant loads include the arterial blood pressure applied to the lumen of the coronary artery, a balloon inflation pressure, and contact interactions between the various components. In the end, expansion of the coronary artery beyond what is generated by a blood pressure load occurs as a passive response to the expansion of a stent, which is facilitated by the inflation of an angioplasty balloon.

Conclusion

It is clear that conventional methods for testing interventional therapies to treat a fibroatheroma can be expensive and time-consuming. FE modeling technologies open new avenues for medical device design engineers to evaluate new concepts for the computer modeling of soft tissues and interventional devices.

Engineers are still working to expand the breadth of information acquired from the soft tissue analyses. There are still a number of questions that remain to be answered. How do various tissue constituents influence the stent expansion process? Will there be a wide expansion because of a particular lipid constituent in the artery? Or, will the stent expand less because of a calcified region? A variety of model enhancements that are in the works will allow further exploration of these questions.  

Copyright ©2004 Medical Device & Diagnostic Industry