Visualizing Cardiac Arrhythmias & Progress Update

Michelle White
Aug 2, 2019 · 9 min read
Image for post
Image for post
Simulated spiral waves. Kaboudian, A., Cherry, E. M., & Fenton, F. H. (2019). Real-time interactive simulations of large-scale systems on personal computers and cell phones: Toward patient-specific heart modeling and other applications. Science advances, 5(3), eaav6019.

The past two weeks of my life were spent at the Georgia Institute of Technology where I learned about the electrical patterns of the heart that result from abnormal heart rhythms, or arrhythmias. Though my time spent in Atlanta feels pretty short in retrospect, I ended up learning a great deal about what causes such arrhythmias and how they are visualized in the experimental setting. Here I will go into detail about the experiments that take place in the CHAOS lab at Georgia Tech.

Heart Experiments

In my most recent post I go into detail about the computer models that I’m working on developing. Such models will eventually be used for simulating the human heart during tachycardia and antitachycardia pacing (ATP) therapy. Up until my visit to Georgia Tech, my experience with cardiac arrhythmias had been purely theoretical and clinical, but I had never seen an actual beating heart in tachycardia or fibrillation. In the CHAOS lab at Georgia Tech they study arrhythmic mechanisms in the hearts of animals ranging from fish to reptiles to mammals.

The most important thing that I learned while watching these experiments was how to visualize arrhythmias from a 3D perspective. Previously, my readings had only prepared me for visualizing arrhythmias from a 2D perspective i.e. the reentrant circuit. I’ll go into more detail about that later, but for now I’ll talk about the rabbit heart experiments that I observed.

The whole purpose of the experiment is to visualize the electrical patterns that cause cardiac arrhythmias. Specifically, we want to see what sorts of mechanisms cause tachycardia and fibrillation in the heart. There are two main stages to the heart experiments: extraction and experimentation. Let’s dig a little deeper into these two stages.

The first stage — extraction — is exactly what it sounds like. The heart that will be experimented on has to be freshly removed from a healthy organism, in this case a rabbit. The heart must be “fresh” because it has to be alive and beating during experimentation — a rabbit heart that is even one day old will no longer be beating and will therefore be unable to show any electrical patterns. The steps to the extraction process are as follows:

  1. A rabbit is anesthetized with ketamine before the extraction takes place so that it feels no pain during the procedure. The rabbit cannot be euthanized because that would cause the heart to stop beating. If the heart stops beating too early, blood will coagulate within its vessels and experimentation would be impossible.
  2. Heparin is injected into a vein in the rabbit’s ear in order to prevent any blood clots from occurring. Heparin is an anticoagulant.
  3. Incisions are made in the rabbit’s chest and the heart it swiftly removed. The remainder of the rabbit is a good source of organs for other groups to study.
  4. The newly-removed heart is flushed with Tyrode’s solution. Tyrode’s solution contains the necessary ions needed for the cardiac action potential, such as sodium, potassium, and calcium. Tyrode’s solution also contains bicarbonate and phosphate as buffers, as well as a sugar as an energy source. In this case we used dextrose.
  5. Tyrode’s solution is circulated through the heart in the direction opposite to blood flow. In other words, Tyrode’s solution is injected through the aorta. This causes the heart to appear turgid. It is important that pressure from the solution isn’t too great, otherwise small blood vessels might pop. Blood vessels need to remain intact so that the solution can reach all parts of the heart and keep it viable for experimentation.
Image for post
Image for post
Diagram showing the direction of blood flow through the heart.

6. Once the heart is sufficiently circulated and all blood is pushed out, it is placed in a cardioplegia solution for transportation. Cardioplegia temporarily stops all activity in the heart (both mechanical and electrical) so that it doesn’t waste energy while in transit.

Now we get to the REALLY interesting part — the actual experimentation. Again, the main point of experimentation is to visualize the electrical activity of the heart during arrhythmia. Some interesting math and techniques are used to accomplish this. The steps for experimentation are as follows:

  1. Once the heart makes it back to the lab, it is removed from cardioplegia and placed in a Petri dish where it receives more Tyrode’s solution. Tyrode’s solution is circulated through the heart and filtered elsewhere to keep it free of particles that could clog blood vessels. The Tyrode’s solution is also kept warm by countercurrent heat exchange. The solution must be kept warm otherwise heart activity will be slowed.
  2. Blebbistatin is given to the heart in order to stop all mechanical activity but allows electrical activity to continue. This ensures that the heart remains still when reading its electrical activity. Blebbistatin is a myosin inhibitor.
  3. A special kind of voltage dye is diffused throughout the heart via the aorta. Again, it is important that even the tiniest blood vessels remain intact so that the voltage dye can reach all parts of the heart. The voltage dye has special properties that cause the heart to fluoresce. Fluorescence occurs when the heart emits light of lower energy than whatever light is incident upon it. In our case, we turned off all lights except for red lights shining directly upon the heart. The heart fluoresced infrared light as a result.
Image for post
Image for post
The electromagnetic spectrum.

5. Fluorescent light is sensed by a camera situated above the heart. For each point on the surface of the heart, a Gaussian surface is read by the camera that corresponds to the wavelength of light fluoresced from that point. Each wavelength corresponds to a different transmembrane potential.

A filter is applied to the Gaussian surfaces and the integral of the filtered portions is taken. The value of the integral is correlated to a specific transmembrane potential and therefore gives us insight into the state of the action potential at every point on the surface of the heart. For this particular voltage dye, the larger the value of the integral, the lower the value of the transmembrane potential. Therefore, when the value of the integral is graphed over several time points, the result is an upside-down action potential. This upside-down action potential is just a mirror image of the actual action potential in the heart.

The computer program used to visualize these electrical patterns 1. reads in what the camera senses 2. applies the filter 3. calculates the integral 4. correlates the integral value with a membrane potential value and 5. color codes the membrane potential values. The end result is a colorful and dynamic visualization of the voltage changes occurring over the surface of the heart with respect to time. While electrical patterns are visualized across the surface of the heart, action potentials can only be visualized at one point at a time.

Image for post
Image for post
Construction of the upside-down action potential. The red line at 630 nm is the filter threshold. The larger the value of the integral, the lower the value of the transmembrane potential. Yu, T. Y., Dehghani, H., Brain, K., Syeda, F., Holmes, A. P., Kirchhof, P., & Fabritz, L. (2017). Optical mapping design for murine atrial electrophysiology. Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, 5(5), 368–376.

6. Once a steady action potential is observed, pacing is delivered from an electrode on the surface of the heart. The entire process is recorded so that we can analyze how pacing affects the arrhythmia. Generally, pacing can terminate tachycardia but not fibrillation.

One drawback of the imaging technique described above is that all images are 2D. If we see spiral waves coming around the edges of an image, we can only assume that there is some type of obstacle that is not within our plane of view. We cannot actually see this obstacle.

You may be wondering what spiral waves even are?? Spiral waves are the primary electrical pattern seen during tachycardia and fibrillation. Spiral waves and reentrant circuits are similar concepts, but not exactly the same. The wave of depolarization that circulates around an anatomical obstacle is visualized as a spiral wave and is exactly the same concept as a reentrant circuit. Examples of anatomical obstacles are scar tissue or burns in the myocardium that are electrically inert and have zero-flux boundaries.

However, there is a completely different type of spiral wave that exists due to functional obstacles. With functional obstacles, the tip of a spiral wave pins to a microscopic heterogeneity in the heart and stays there unless disrupted by ATP or far-field pacing. Functional obstacles, or microheterogeneities, can come in the form of fibroblast cells, clusters of dead myocytes, or misaligned gap junctions, among other things. If tachycardia starts due to reentry around an anatomical obstacle, tributaries from the anatomical spiral wave can pin to nearby microheterogeneities and create a bunch of smaller functional spiral waves. This is what happens when tachycardia degenerates into fibrillation.

Image for post
Image for post
Examples of microheterogeneities. A: myocyte-free regions; B: mechanical damage from forceps; C: areas of higher density; D: clusters of damaged cells; E: areas of lesser/uneven density. Agladze, K., Kay, M. W., Krinsky, V., & Sarvazyan, N. (2007). Interaction between spiral and paced waves in cardiac tissue. American Journal of Physiology-Heart and Circulatory Physiology, 293(1), H503-H513.

Spiral waves interact with each other in many different ways. When spiral waves crash, they can either terminate or combine to make a bigger spiral wave. There is lots of research being done currently to study the effects of ATP on functional spiral waves. The goal of therapeutic pacing is to unpin the spiral wave and push it away from its functional obstacle. Hopefully then it will either terminate due to instability or due to collision with a zero-flux boundary.

From a 2D perspective, a functional obstacle can simply be thought of as a point. However, from a 3D perspective, a functional obstacle becomes a filament. Trying to decipher the electrical patterns of a heart during fibrillation becomes much easier when keeping in mind that spiral waves can originate from within the walls of the heart or from behind the plane of view. Heart muscle is anisotropic so it has lots of potential for heterogeneities to crop up and serve as anchors for spiral waves.

Updates & August Plans

Further developments in the MATLAB code that I discussed in my most recent article include sliding-window detection and re-detection of ventricular tachycardia (VT). If the code senses a cycle length below some threshold, it begins sliding window detection over the next 10 beats. In other words, an array stores 2 cycle length values at a time. With each time step, it drops the previous value and adds the next, so that at any given time the window can only see 2 values. A separate variable keeps track of how many of those values are below the threshold.

If at least 8 out of 10 of those values are below the threshold, then my detection script tells my ATP script to deliver ATP. There is also a re-detection phase after ATP is finished to confirm if tachycardia is terminated or not. If it isn’t terminated, more ATP is delivered. After more unsuccessful ATP the code defaults to shock. Right now I can only set my ATP algorithm to asynchronous burst pacing, but I plan to incorporate different basic ATP and AATP algorithms in the future.

Image for post
Image for post
MATLAB output showing a 4-impulse ATP burst (top graph) being delivered after 10 tachycardia beats. In this case, all 10 detected VT cycle lengths were below the 600 ms threshold. ATP successfully terminates tachycardia and normal sinus rhythm ensues. SPA = slow path antegrade; SPR = slow path retrograde; FPA = fast path antegrade; FPR = fast path retrograde

I am also working on learning OpenGL and GLSL (GL Shading Language) in order to program and simulate the electrical patterns and spiral waves I mentioned earlier in this article. Having experimental data + simulated data will enable us to make a clinically-relevant heart simulation. The goal is to simulate spiral waves and generate synthetic electrograms (EGMs) from them. Then, we will use machine learning to determine the characteristics of the spiral waves based solely off of their EGMs. The EGMs will then be fed into my ATP emulator, possibly with the help of a convolutional neural network, prompting it to deliver ATP according to the characteristics of the spiral waves. Thus another aspect of this project is optimizing ATP algorithms when given specific EGM/spiral wave characteristics.

Lastly, I would like to give a big thank you to the members of the CHAOS lab at Georgia Tech. To learn more about the CHAOS lab, please visit their site at the following link!

Medical CPS

Research, ideas, knowledge about medical cyber-physical…

Medium is an open platform where 170 million readers come to find insightful and dynamic thinking. Here, expert and undiscovered voices alike dive into the heart of any topic and bring new ideas to the surface. Learn more

Follow the writers, publications, and topics that matter to you, and you’ll see them on your homepage and in your inbox. Explore

If you have a story to tell, knowledge to share, or a perspective to offer — welcome home. It’s easy and free to post your thinking on any topic. Write on Medium

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store