A Patient-based Model to Investigate the Effect of Micro-structure Scars on the Optogenetic Defibrillation Success: A Computational Modelling Approach

This study investigates the role of the microstructure of real scars in the success of optogenetic defibrillation. To reduce the computational cost of high-order models (like Ten Tusscher Model, TTM) for a single cell as well as to take advantage of their ability to generate a more realistic output, we developed a low-order model of optogenetic cardiac tissue based on the modified Alieve-Panfilov single-cell model and estimated its parameters using a TTM. Two-dimensional electrophysiological cardiac tissue models were produced including different scar shapes that were extracted from Late Gadolinium-Enhanced (LGE) magnetic resonance imaging data set of 10 patients with non-ischemic dilated cardiomyopathy. The scar shapes were classified based on four criteria: transmurality, relative area, scar entropy, and interface length. Scar with the highest 25% of the relative area showed 25% of successful cases, this ratio is 27%, and 25% for a scar with the most top 25% of entropy, and transmurality, respectively. In comparison, the proportions are 61.54%, 44.44%, and 61.76%, for the lowest 25% of the area, entropy, and transmurality. We also investigated the efficacy of various methods for light-sensitive cells’ distribution within the cardiac tissue with scar. Four types of distributions were defined. Defibrillation within tissues with 0.1 light-sensitive out of all cells was 15 to 25% more successful than their counterparts with 0.05 light-sensitive cells. Lastly, we examined the effect of an earlier stimulation on the success probability of defibrillation. Our results indicated that inducing 0.5 msec earlier resulted in a roughly 15% rise in successful cases.


42
Ventricular fibrillation is one of the most prevalent heart arrhythmias, which might result in blood clots, stroke, or even in serious 43 conditions, heart failure. As a conventional treatment, an implantable cardioverter-defibrillator (ICD) is used for a heart prone to 44 arrhythmia. Unfortunately, ICDs have some undesirable side effects, including structural damage to cardiac tissue, which, in turn, 45 would result in a more heterogenous cardiac tissue. These side effects have inclined scientists to find new medical approaches 46 addressing fibrillation problems. Optogenetic treatments (defibrillators) are convenient candidates to substitute for cardiac therapies based on electrical stimulations of the cardiac tissue [1,2].

48
In 2005, Li Li et al. examined the mechanism of shock-induced arrhythmogenesis and arrhythmia maintenance in a rabbit model of 49 a healed myocardial infarction. They proved that an infarcted rabbit heart is more vulnerable to shock-induced arrhythmogenesis 50 and arrhythmia maintenance [3]. Also, in another research [4], utilizing MRI and DTI data, researchers developed a 3-D model of the infarcted canine heart, by which they claimed that borders between intact and scar tissues are the location from which subsequent 52 fibrillations will start, because of their resulting heterogeneity. Regarding the fact that a considerable number of candidates for ICD 53 implantation are who have a prior myocardial infarction, the evaluation of the efficacy of treatments in such conditions would be 54 crucial. In [5], a 3-D model was developed, and then the authors showed that electrical defibrillators have lower chances of being 55 successful in infarcted hearts. According to the difference between the electrical and optical stimulations, understanding the effect 56 of optical defibrillating shock within an infarcted cardiac tissue is highly essential.

57
In this study, we developed a low-order computational model using LGE_MRI data of non-ischemic dilated cardiomyopathy 58 patients. Utilizing this model, we sought to investigate the efficacy of optical defibrillation, while investigating the effect of the 59 fibrosis microstructure and also the distribution of light-sensitive cells through cardiac tissue. First, we simulated the action potential 60 of normal and optogenetic single cardiac cells using the Ten Tusscher cardiac cell model (TTM it is precise enough for describing cardiac fibrillation and defibrillation procedure, since in such a phenomenon, the exact shape of 101 the action potential is not needed, and just passing the activation threshold is the issue. Therefore, in order to take advantage of the 102 Ten Teusscher model, using the output of the resulting Ten Tusscher action potential for both light-sensitive and normal cells, we 103 estimated parameters of a low-order cardiac model, the Aliev-Panfilov, in a way that it approximates the timing of Ten Tusscher 104 action potential as well as possible.
107 113 As discussed in [11], the parameter "a" is called the threshold; the parameter " " describes the ratio of the time scales of variables u 114 and v.

115
The Aliev-Panfilov model, because of the uv term, which shapes the nullclines geometry, is more appropriate for describing the 116 cardiac cell dynamics than the Fitzhugh-Nagumo model. In this study, we changed (2) to (4) [12] 117 Where b = 0.05, i.e. the parameter "a" in (1) and (2)   can be observed that, compared to AP action potential shown in Fig. 2, that of Fig. 3 is much more similar to the TT model's AP. We sought to simulate the defibrillation procedure, so solving the bi-domain process was a crucial part of the study; since it can get 147 us access to intracellular and extracellular spaces, independently. The bidomain problem equations are as follows. 148 149 We utilized the five-stencil finite difference approach to solving these equations. Finally, the resultant equations are as follows.   198 We defined successful defibrillation as a stimulation which scilences (damps) entirely the spiral waves. Based on this definition, 199 the percentage of successful defibrillations among all cases was calculated and shown in Table 2. As clearly can be seen, the DLPH 200 distribution of light-sensitive cells has a considerably lower amount of fibrillations being completely silenced. As discussed in Materials and methods, for each scar shape extracting from LGE-MRI data, we calculated the amount of four 203 features. Also, we divided each feature into four quartiles, namely 1 st , 2 nd , 3 rd , and 4 th quartile. Table 3 shows the percentage of

240
Examining other choices for P and D in future studies, will result in a better view for understanding the role of distribution of the 241 light sensitive cells in the success/ failure of defibrillation procedure via optogenetic methods.

242
As the last part of this study, we induced a second optical stimulation at time step 15000 (i.e., 1.5 msec after the first stimulation) 243 for the DLPL group to investigate the role of primary (former) stimulation. We observed that the percentage of successful cases 244 after a quick stimulation is higher than when one single stimulation is applied later. Our results show a roughly 15% rise in the 245 successful cases. The point at issue is that the time of stimulation could have an important role in the success of defibrillation.

247
There are two issues to be considered in this study. First, we utilized a simplified model for each cell, and second, our study was 248 performed in a 2-D tissue model. Both simplifications were accepted to help us to reduce the cost of computations. Therefore, our 249 results can be considered as preliminary results giuding us towards the main answers. As our future work, we will extend the