M. Wilhelms. Analysis of cardiac ischemia regarding ECG and stability in a computer model of the human ventricles. Institute of Biomedical Engineering, Karlsruhe Institute of Technology (KIT). Diplomarbeit. 2009
Acute cardiac ischemia is characterized by a deficient supply of the heart muscle, caused by the occlusion of one or more coronary arteries. In the temporal course of this pathology, mainly three effects can be observed: hyperkalemia, acidosis, and hypoxia. Consequently, electrophysiological characteristics, as e.g. the APD and Vm, are modified heterogeneously depending on the distance to the occlusion site. These temporal and spatial heterogeneities cause changes of the ECG and the excitation propagation pattern, which was analyzed in this work.An existing description of ischemia effects on the cellular level was used for tissue simulations. At first, a three-dimensional ventricular model was adapted, so that the spatial heterogeneities presenting during cardiac ischemia could be simulated as realistic as possible. After initializing the effects of ischemia in a single-cell environment until the requested stage was reached, differ- ent ischemia configurations were simulated. For this purpose, the transmural extent, the relation between CIZ to BZ, the size of the ischemic region, and the location of the occluded coronary ar- teries were varied. Afterwards, the corresponding ECGs were forward calculated. Here, significant changes of the ST segment could be observed. Depending on the occlusion site, the leads placed above the ischemic region were affected most by this pathology. In case of subendocardial ischemia, a ST depression in leads above the ischemic region can be noticed, whereas transmural ischemia leads to ST elevation there and ST depression at the border to healthy tissue. Depending on the relation between CIZ and BZ, the beginning of the repolarization (onset T wave) is delayed. If the BZ is large, the repolarization starts earlier than if the BZ is small. By this way, the length of the ST segment is influenced. The size of the ischemic region also influences the voltage decrease in the ST segment. Larger ischemic regions cause a stronger decrease.The temporal and spatial heterogeneities occurring during this pathology are known to favor reentry . Therefore, the next part of this diploma thesis dealt with the investigation of reentrant circuits initiated during cardiac ischemia. For this purpose, a two-dimensional tissue model with a circular ischemic region, as introduced in , was used. However, the initiation of stable figure-of- eight reentry was difficult due to the strongly varying effective refractory periods in the different ischemic regions. Therefore, different combinations of ischemia stages and coupling intervals were tested. Furthermore, the CV was too high and the APD of human cells too long for this geometry, so that different types of blocks (e.g. at the BZ or CIZ) occurred, as refractory tissue was reached. As a consequence, only one reentrant circuit could be initiated under these conditions. For that reason, the CV was reduced from 800mm/s to 500mm/s, as shown in . By this, a stable figure-of-eight reentry was initiated in this case. Finally, a numerical method for the solution of the monodomain equation was implemented in the existing simulation environment acCELLerate. Generally, this semi-implicit solver requires the solution of an additional linear system of equations. Thus, it is slower than the explicit solver, if the same time step is used. However, the semi-implicit solver is more stable and accurate than the explicit one, especially at high spatial resolutions. Therefore, longer time steps can be used, achieving the same accuracy as the explicit solver. Consequently, the total computing time can be reduced depending on the cell model by using the semi-implicit solver. However, this effect is present only, if the additional calculations contribute to less extent to the total computing time. Thus, the cell model has to be computationally complex, so that the application of the semi-implicit solver reduces the computing time.