The 3D Heart

By Tim Hornyak

How a computer simulation of the aorta could help surgeons make better, quicker decisions.

It helps to check a map before starting a road trip, so imagine how important it is to know the lay of the land, so to speak, before you crack open someone’s chest. That’s exactly what doctors do to prepare for heart surgery: They use magnetic resonance imaging (MRI) and computed tomography (CT) scans to look at what’s inside the patient before making the first incision. The technology, though, is far from perfect. Unlike the precise satellite images we use to navigate city streets, these heart images are often noisy and tricky to interpret. But what if doctors could have a 3D map of the aorta before the patient goes under the knife? A team of McGill researchers is collaborating with an industrial partner to develop technology that could become an essential decision-making tool for cardiac surgeons.

Heart disease remains one of the top killers of Canadians, and is a particular threat to older people. Roughly one-third of all heart surgeries are condi- tions involving the heart’s valves. One such ailment, aortic valve disease, affects the gateway to the body’s main blood vessel. It occurs when the valves regulating blood flow from the left ventricle into the aorta become compromised due to bulges in the aorta wall and other complications. To fix the problem, surgeons may dilate or implant grafts in the ascending aorta, a conduit just above the heart that’s shaped somewhat like an umbrella handle.

Rosaire Mongrain of the Department of Mechanical Engineering points toward a colourful 3D mesh on his computer screen. It looks like a wire-frame tree trunk with a thick stalk growing out of a bulbous base. This is a model of an actual ascending aorta, based on MRI and CT scan data. When Mongrain clicks his mouse, the tree begins to pulsate with an invisible liquid. The model is replicating the function of the aortic root.

“We are extracting the 3D anatomical shape of a patient’s heart and we reconstruct it on the computer and then assign mechanical properties to it,” Mongrain explains. “The surgeon may replace some parts of the aorta with artificial structures. These grafts do not have the same properties as the natural structures, and the surgeon will want to understand the impact on the blood flow.”

“The software gives doctors some options to think about prior to surgery,” says collaborator Richard Leask of the Department of Chemical Engineering. “We can tell them, ‘OK, if you put that coronary ostia up at this level, this is how much flow you are going to get.’ Before they go into the operating room, they can get an idea of how they will do the surgery based on the actual flow parameters, whereas up until now they just do it based on an image. Ultimately, the idea is to improve the post- surgical outcome of these patients.”

Mongrain and Leask have been working with Raymond Cartier, a heart surgeon at the Montreal Heart Institute. Inspired by Cartier’s work on aorta surgery and seeking to better understand the flow features of the aorta, Mongrain and Leask began the project six years ago and have gradually made their model more sophisticated. There are other efforts to produce computer models of the human body, such as the large-scale European HUMOS project to create a “numerical human” for applications such as car crash simulations, but few are as tissue-specific as the McGill biomechanical model of the aorta. In 2010, Mongrain, Leask and Cartier received a three-year grant from the Natural Sciences and Engineering Research Council of Canada to build the aorta model.

The researchers have shown that their basic concept is valid. They have taken about 120 aorta samples from cadavers or surgeries, and tested them for flexibility and other mechanical and biochemical properties. Then, they replicated them as CAD models using their software as well as commercial simulation products such as Comsol, LS-DYNA and 3D-Doctor. To verify the validity of a model, they send its CAD file to a rapid prototyping machine to manufacture a life-sized mockup of the aorta in transparent elastic silicone, complete with sinuses, arch and upper branches. Using a technique called particle image velocimetry, the team can then check the flow parameters of a fluid coursing through the ersatz aorta by using laser light to reflect off titanium oxide particles. Though there were some differences between the silicone model and flow results from the actual organ using echocardiography, the pair says they should diminish as the software is improved.

“Computationally, these simulations are very intense, and they can take weeks to converge,” says Leask. “To get good results you need a very fine mesh, meaning millions of nodes. But we hope to get the computation time down to a number of hours or to a day.” To achieve that, the re- searchers will have to develop tools to automate the process of turning the raw imaging data into a useful graphical model of the aorta. They will also see how much the resolution of the model can be reduced, so it can be rendered more quickly without affecting its usefulness as a guide.

Since aortic repair surgeries are elective, surgeons have time to plan the operation with X-rays, MRI and other means. Running a scan of the aorta and then simulating how it will behave if a certain graft or valve is implanted would be part of pre-surgical preparation. Mongrain even envisages his aorta model as a standard feature on imaging equipment that shows various surgical possibilities to doctors.

“For example, on a GE MRI machine you could have a little subroutine on the console with the command ‘simulate surgery’ and it starts to generate the model and the flow,” says Mongrain. “It’s a little like Star Trek.”

The researchers plan to make the simulated aorta available to two types of end users in two years. One type is surgeons. They may require some assistance from engineers in interpreting certain data, but the system would have a simple graphical user interface as a surgical planning tool. The other end user is corpor- ations. Coroneo, a private-sector biomedical company in Montreal, is the industrial partner in the research project. It wants to use the software model in the design of its devices such as aortic annuloplasty rings, which are placed around the aorta to prevent aneurysms.

The main technical problems with developing the model have been resolved. Refining the system, and incorporating various pathologies of the aorta to make it increasingly realistic, is the next step. While corporate users will want a more unrefined version of the system so they can hack around with it, the team is now focused on making the software more user-friendly, polished and smooth for surgeons.

“We need to automate the system for medical doctors,” says Leask. “Heart surgeons have no time to spare. They have many patients and no patience.”