Blog Archives

Reflection

Anatomy of the heart

When we started the project we had very limited knowledge of the anatomy of the human heart. Therefore our first order of business was to learn more about all the different parts of the heart, what their function is and about various congenital heart defects. The cardiologists in Leiden helped us with this by showing us several preserved hearts and explaining the basics of how the heart works. They also provided us with some online courses from which we could learn even more. This took up most of the first week. At the same time we took a look at the segmented CT-scan of a patient’s heart that they had already prepared for us, so we could see if we could work with it.

 

From CT to CAD

At this point our knowledge of CAD-software and file types that could be used to create a printable model was very limited. Especially working with meshes was something we had little to no experience with. Because the file created with the CT-scan was a mesh, we spent quite some time researching other options. We looked for ways to convert a mesh into a NURBS model, we searched for existing models that we could adjust and change and even tried to model a heart from scratch. All with varying but not very promising results. Converting a mesh didn’t work at all or resulted in a model that was difficult to work with, as were most of the existing models we found. Creating a heart from scratch seemed possible but was very time consuming.

But then we had a breakthrough. We found a piece of software that allowed us to easily adjust the mesh models from the CT-scans. Working with the model from the scan would mean a higher anatomical accuracy. To get the computer model as close to the scanned heart as possible, Leiden scanned one of the preserved hearts for us. Because a preserved heart is not beating and inside a body anymore it is possible to use more radiation and get a higher resolution. However, we could not get everything out of the scan. The valves were too thin to see, so we had to model them ourselves.

 

Progress in the process

At this point everyone in the group started to get their own place in the process and develop their own specialty. This prevented us from doing any double work and made it possible to learn new software and techniques in a very short time. We had someone who cleaned up the mesh model and made it into a solid, someone who modeled the coronary arteries, someone who modeled the valves and someone who put everything together into a correct, printable model. During this process we also started making our first test prints, which were mostly valves, to see what the materials were really like and how thin, soft or hard we could make everything.

 

Printing

It ended up taking longer than we initially thought and wanted to start printing our first entire heart. This was due to the fact that it took us a long time to find the appropriate software, learn how to use it and also combining all the separate parts we had into one printable solid. However, because we did take our time for this process we ended up with a set of steps that were easy to repeat. So once we had our first printed heart and the cardiologists from LUMC gave us some very useful feedback about the anatomy of it, we were able to improve the heart quite fast. Not only did we adjust it anatomically, we also gave various parts of the heart different colors and material properties. We did this with the purpose of making it easier for patients to understand the model.

In the end we had a printed model of a healthy heart, without defects. The colors of the model were not exactly what we wanted yet, due to it being too late to switch the colors in the printer. However the different colors that we did use gave a good impression of what is possible. The different material settings also turned out well and both could be easily adjusted for a next print if necessary. The valves could anatomically still be located a bit better and maybe the heart could also be cut in a different location, or more than one, to make them more visible. But with the process we created and our skills that we are still developing as we go, this could all be easily changed.

 

Up next

The next step would be to create a model of a heart with a defect. The same methods as we used for a normal heart can be used for this. Fitting in the modeled parts in an anatomically correct way will probably be a bit more challenging, because the defect can make the heart look very different. This makes it more difficult to find the right location of, for instance, the valves. However, with the expertise of the cardiologists at LUMC it is definitely possible.

No heartstrings attached

Since we were using the CT-scans that the LUMC cardiologists provided us with, we had a model of the heart that was more or less anatomically correct. However, the valves were missing from this model. The scan had a resolution with voxels of approximately 1 by 1 millimeter. Because the cusps of the valves and the heart strings are thinner than that, they didn’t show up on the CT-scan. Therefore we had to model them ourselves.

The aortic and pulmonary valves look quite similar. They are both semilunar valves and consist of three leaflets that are shaped like a half full moon, hence the name. These were fairly simple to model in Rhino. We did a test print to see what material combination we could or should use and to show in Leiden. We got some useful feedback. On the first try the connections of the cusps with the artery were located in a single plane, while in reality the leaflets are attached to the arteries in semi-circles. This was quite easily fixed so then the aortic and pulmonary valve could be placed inside the heart. Because the aorta and the pulmonary artery always roughly have the shape of a circle, this model can be used in almost any heart.

Aortakleppen

The tricuspid and mitral valve were more of a challenge. They are also quite similar, except for one major difference: the tricuspid valve has three cusps and the mitral has two. Therefore the modeling principles could be used for both, but they did result in two separate models. What made it difficult to model were the very thin heart strings that protrude from the edges of the leaflets and connect them with the papillary muscles that are attached to the walls of the ventricles. Every heart is shaped a little differently, especially ones with a cardiac defect. For that reason it was vital that the valve would be easy to adjust and the heart strings could be deformed and their ends dragged to connect to different places. At the same time we had to make sure the cusps and strings were thick enough to print properly. Our test piece showed that the strings and the leaflets required a minimal thickness of 1 millimeter and probably more to increase the durability. Anatomically this would not be entirely correct, but it would still look small in relation to the rest of the heart and show how the connection is made. In Rhino we modeled the leaflets as one piece, because in a real heart they are all connected as well. The strings were created separately and made adjustable with Grasshopper, which we also used to automatically combine all the parts into a closed polysurface so that the valve would be printable.

MitralisTricuspidalis

In the end we had three separate models. One that can be used as the aortic and pulmonary valve, one tricuspid valve and one mitral valve. The semilunar valve can be placed in the heart model by simply scaling and rotating it. The ventricular valves can be put in the right position by deforming the cusps and dragging the heart strings to the right place.