Abstract:

Cardiac arrhythmia like atrial flutter is a threat to the human health. Often the cause of theillness cannot be determined precisely. The treatment is difficult and often combined withintervention like radio frequency catheter ablation where diseased tissue on the atrium iselectrically isolated by a lesion induced by the heated tip of the ablation catheter. An electrophysiological examination precedes with the help of different kinds of mapping catheters.Electrodes at the end of those mapping catheters are capable to detect cardiac excitations andsave the temporal and spatial information in local activation time maps which serve as theinput for the methods used in this thesis. Especially slow conducting areas are consideredarrhythmic and may cause the looping excitation to persist and can therefore be the source ofthe arrhythmia. The conduction velocity describes the direction and the magnitude of thepropagation of the wave fronts of the electrical excitation leading to the muscle contractionof the human heart. In that matter the determination of the conduction velocity (CV) plays afundamental role in the process of placing lesions in arrhythmogenic tissue thus being ableto save healthy areas in the process. Additionally, the success rate of the procedure may beimproved by helping the treating physician to locate all possible areas for the cause of thearrhythmia which otherwise might be missed. This thesis will explain and evaluate two majorapproaches for the estimation of the conduction velocity for both simulated and clinicaldata sets. With the help of trigonometric functions like the law of cosine the triangulationapproach is capable to estimate the CV between three points. The polynomial fit makes useof the least squares approach and selects viable points for the estimation of the CV in a smallarea around a point of interest. Under the influence of noise added to simulated data bothalgorithms are statistically compared in consideration of the robustness and the accuracy ofthe results. Two clinical data sets will be analysed and the results are interpreted towards thequestion if they are applicable in medical environment. In simulated data under the influenceof noise up to a certain level the triangulation approach yields conduction velocity maps(CV maps) which show local heterogeneities but is not capable to determine the exact CVeven in noise free data. The results of the polynomial fit are very close to the ground truthvalue for the CV in simulated data but suffer heavily under the influence of noise. In clinicaldata both algorithms are capable to locate slow conducting tissue and display the excitationdirection of the propagation. The polynomial fit approach gives the propagation directioneven in very noisy data whereas the triangulation loses this capability when the noise levelsurpasses a certain value. The triangulation seems to yield results with higher resolution andborders of possibly arrhythmogenic tissue patches can be determined more precisely.