1. Physically-based models for Catheter, Guidewire and Stent simulation
Julien Lenoir
Stephane Cotin, Christian Duriez and Paul Neumann
The SIM Group – CIMIT, MGH, Harvard Medical School
Good morning. I am going to present some of our work on developing a suite of simulated interventional radiology devices for a computer-based training system.

2. Interventional Radiology
The context of our work in Interventional radiology which applies therapies through the human vascular system with a series of wire like devices. Real-time navigation is accomplished through dynamic X-ray imaging and injecting of contrast agent into the blood stream. In neuroradiology, an interventionalist would enter in an artery at the patient’s hip, navigate a catheter/guidewire through the abdomen pass the heart in through the neck then finally into the brain. They would study flow pattern to detect any anomalies such as blockages, aneurysms, or AVS and then apply the appropriate treatment.
Providing therapy through the
human vascular system to
prevent stroke.

3. Interventional Devices
Primary Devices: guidewire and catheter
Attributes: Flexible, smooth, uncompressible
Manipulate: Insert/retract and twist externally
Navigation: curved tip for vessel junctions
Our Goal: To replicate these
devices within a simulator to
aid in training.
The context of our work in Interventional radiology which applies therapies through the human vascular system with a series of wire like devices. Real-time navigation is accomplished through dynamic X-ray imaging and injecting of contrast agent into the blood stream. In neuroradiology, an interventionalist would enter in an artery at the patient’s hip, navigate a catheter/guidewire through the abdomen pass the heart in through the neck then finally into the brain. They would study flow pattern to detect any anomalies such as blockages, aneurysms, or AVS and then apply the appropriate treatment.

4. Challenge Thin Stiff Structures Nested Devices
Multiple Contacts and Sliding conditions
Large Deformations
Requires High Fidelity for Radiologists
The context of our work in Interventional radiology which applies therapies through the human vascular system with a series of wire like devices. Real-time navigation is accomplished through dynamic X-ray imaging and injecting of contrast agent into the blood stream. In neuroradiology, an interventionalist would enter in an artery at the patient’s hip, navigate a catheter/guidewire through the abdomen pass the heart in through the neck then finally into the brain. They would study flow pattern to detect any anomalies such as blockages, aneurysms, or AVS and then apply the appropriate treatment.

5. Previous Work General one-dimensional models
Dynamic spline [Lenoir et al 02, Nocent & Remion 01]
Static Cosserat model [Pai 02]
Specific catheter simulation
Rigid bodies and joints (multi-body dynamics) [Dawson et al 00]
Linear elastic FEM [Nowinski 01]
Incremental FEM model [Cotin et al 05]
In replicating this type of procedure within a computer-based training simulator, we need to be able to compute the physical behavior of these wire-like devices. Previous work from the graphics community has to use Splines and multi-body dynamics models, while the Cosserat model and FEM algorithms have been put forth from the mechanical community. Our current approach is an incremental FEM model with a sub-structure analysis.

6. Physics-based Representation
Base Model
6 Degrees of Freedom (translation + rotation)
Linear elasticity
Optimization
Incremental FEM for
geometric nonlinearity
Performance improvement through
sub-structure analysis
Other details about our approach. Each beam element/node has six degree of freedom, static linear, substructure analysis. Besides the physics, another challenge is resolving collisions since these type of devices are in near constant contact with the vessel walls. Resolving collisions involves detection followed by response.

7. Anatomical model is described in
Collision Detection
Accomplished through optimized vascular model
Oriented graph
Each beam node of a device is tracked using
Proximity measure
Temporal coherence
Surface Partitioned
return active section
test local triangle subset
To determine if a beam node of our device has penetrated the vessel wall, we need to quickly determine the violating surface element. This is accomplished by an optimized vascular model which has a simple framework and its surface which is partitioned. We can quickly determine which segment a bean node is within an test only its surface elements. This optimized anatomical model is pre-computed and fast to maintain at run-time.
Anatomical model is described in
another paper on pp

8. Collision Response
Collision response needs to account for multiple contacts and sliding conditions
Quadratic Programming proved to be too time consuming
Our Approach: Iterative Gauss Seidel
Use penalty method locally
Propagate change to other nodes
Iterate checking other nodes until no violations
Collision response is more tricky. Simple penalty methods cnnot be used. Quadratic programming proved to be too time consuming. Our current approach is a Guass Seidel iterative function which constraints are broken into convex and concave sets then the device nodes are subdivided into small groups with additional boundary constraints. Then these merged and their displacements for the bean nodes are computed. This cycle iterates until convergence.

9. Results

10. Results

11. Co-axial Catheter/Guidewire
One unified device rather than two
Modulate material properties based on regions:
Overlapping
Guidewire only
Catheter only
Locally update material properties using Halpin-Tsai equations.
Our first example models the most common device state within a procedure – a nested guidewire within a catheter. Instead of modeling two separate devices with constant contact, we model them as single model with a function which moduates material properties in separated regions and overlaps regions. In figures you can see how the nested guidewire deforms the its co-axial catheter which is necessary for navigation.

12. Results – Video 2
This is a simple video which illustrates this interaction between the two devices.

13. Stents Thin cylindrical metallic mesh
Implanted to open partially blocked vessels restoring blood flow
Expands radially when released
Moving into theuraptic devices, we can apply our physics-based representation to stents which are a expanding metallic mesh to increase ia vessel’s flow. Here again we modeling a skin surface which expands and collides with a vessel’s surface. We currently designate subset of skin points which we will test for collision for speed. We’ve set up a relationship such forces are propagated from the skin surface to the underlying beam node model and visa versa.

14. Stents Modeling Issues:
Connect an additional surface to ‘core’ beam model
Surface constrained and expands when released
Perform collision test on surface elements
Coupled relationship between surface and beam nodes
Forces on surface elements propagated to beam nodes
Currently working on local vessel deformation
Moving into theuraptic devices, we can apply our physics-based representation to stents which are a expanding metallic mesh to increase ia vessel’s flow. Here again we modeling a skin surface which expands and collides with a vessel’s surface. We currently designate subset of skin points which we will test for collision for speed. We’ve set up a relationship such forces are propagated from the skin surface to the underlying beam node model and visa versa.
Surface Elements
Beam nodes

15. Stent Video
This video shows a stent being deployed with our simulation system. We’ve created a stenosis within one of the cardioids. Contrast injections were determine stenosis to interventionalist. Place the stent with our catheter. Release the stent. This video was recorded in real-time.

16. Conclusions and Ongoing Work
Real-time robust physics-based representation for wire-like devices
Defined composite technique of nested devices
Demonstrated extensions: stents
Ongoing work
Local and global deformation of anatomical model
Develop angioplasty balloon using similar principles
Investigating coils
Finalize complete simulation system
In conclusion, we have present a robust physics-based representation for simulating wire-like devices such as guidewires, catheters, stents and soon balloons. All device computation is within interactive rates. We currently implementing global and local deformation of a vascular model. When this is working, we believe that implementing angioplasty ballons devices should be straight forward. We’re also investigating to see if our model can be applied to coils which are used to fill aneurysms. Lastly we are polishing our current simulation system.

17. Acknowledgements – Team – Xunlei Wu
Vincent Luboz Julien Lenoir Christian Duriez Paul Neumann Stephane Cotin
– Funding –
TATRC CIMIT
Finally, here is our entire team who are working on this project from the SIM group. I also want to thanks our funding source, TATRC and CIMIT, for enabling this work to be carried out. Very last, I would like to thank you for your attention.