InitRech 2015/2016, sujet 11 : Différence entre versions

Version du 18 juin 2016 à 15:21

Summary

When a patient suffers cardiac arrhythmias or heart failure, caused by an abnormal electrical activity in the myocardium, the standard procedure is to locate where the problem come from, and then destroy the cells that creates the unusual heart-beating by heating them using radio frequency ablation.

In order to train rookie surgeons on this kind of surgery, HAL proposed an interactive training system that allows the surgeons to practice based on a software called SOFA. Even if system like that already exist, the one introduce in this article (based on a device that already exist) combines a biophysical modeling of cardiac electrophysiology with an endovascular catheterization. The simulating framework allows to model a mapping of the heart, to do the ablation with real-time feedback, and then gives a synthesis of the surgery.

Based on an European database, a 3D view of the endovascular navigation around the heart is first built, using image visualization framework and software. Then, a venial view is also made, using a data set from a company. To finish, they simulate the heart beating by using an algorithm, which provides in the end the cardiac motion.

The GPU used in the program is based on Mitchell Schaeffer model. They choose this model because all 5 parameters had a physiological meaning, and the simulation is more acurate with this method. It captures the potential V_M of the transmembrane. In order to improve the result of the computer, they choose to apply this simulation only on the ventricules. They explain that it can be done because ventricles and atria are electrically isolated. With more than 30k elements, the electrophysiology of the ventricle is computed on a mesh. The time step used on this part is 10^-4 s. To do every simulations at the same time, they use threads on the CPU with a scheduling structure : one thread for endovascular navigation and electrophysiology simulations, and another thread for the graphic view of the simulation. Both thread are asynchronous, because the time step for navigation simulation is greater than the time step mentioned before (0.02s & 10^-4 s).

To locate the pathological area, the surgeon uses a system that detect the zone with the fastest activation. It's the first step to do before strating the operation. Catheters, used to mesure a potential (or a difference of potential), give a tool for the cardiologist to interpret the electrical activity of the heart. Mesure a difference of potential is better, but this can only capture a potential.

As the surgeon mainly rely on the electrophysical signals, it's really important for the model to be trusted and precise. As so, they made a fonctionality that gives the bipolar signal by doing the difference between two unipolar signal, which gives a better precision.

The mapping evolves as the catheters go on the heart. For exemple, if one touches the heart wall, the mapping will change this way, and some triangles are displayed to warn the user. When he reaches the region to operate, the cardiologist uses an AC RF signal (between 300 and 700 kHz), which is delivered in an electrode in contact with the tissue to remove. To know if the operation was a succes, the electrical conductivity of the operated area must be none. The simulation modelize that by decreasing progressively the electrical potential of the area. The trainee can also choose the power of ablation, so the operation can be faster or slower. Then, to check if the operation was a succes, the simulator inject current (by pressing a pedal) to the area, and if no abnormal activity are dectecting, it means that the ablation was a success.


To do such a difficult and precise work, the simulator needs good hardware : i7 processor, NVidia GTX 580 GPU. A table of result reveals that to be most effictive, the rookie surgeon needs to touch the heart wall the less possible time.

As a conclusion, we can say that this simulator is really close to real time, but still needs some test in clinic to asses the model.

Main contribution

The aim of the article is to introduce a system that can simulate a specific heart operation.
The main contribution of the project is about helping the rookie cardiologist to improve their skills, by having them training on a system that can tell them if they have done some mistkaes. It present a new way of training, better than the old one (who was to train on a real patient, with the supervision of a senior cardiologist). Even if a real practice is always good, it's better to pratice first on a simulation program, which wont hurt anyone.
The authors explain how every single part of the simulator work, and how they are connected to each other. They also show the results of the simulation, and point why and how the simulation is sometime worst than in an other time.

Applications