Atrial fibrillation (AF) is a cardiac arrhythmia that is becoming a growing public health problem. The most widely used minimally invasive method for treating AF is radiofrequency ablation (RFA), in which a high-frequency alternating current is delivered into the tissue via the tip of an ablation catheter. The release of energy in the form of heat at the contact with cardiac tissue leads to permanent localized damage to the tissue (lesion) at temperatures above 50 °C, thus permanently disrupting the conduction of the action potentials in the ablated area. The basic idea of radiofrequency ablation for the treatment of atrial fibrillation is to induce a continuous line of localized lesions into the tissue around the entrance of the pulmonary veins to the left atrium. This creates an ablation line that electrically isolates the pulmonary veins from the left atrium and thus also eliminates the triggers of atrial fibrillation, which are located in the pulmonary veins in approximately 90% of cases. This procedure is also called pulmonary vein isolation.
During radiofrequency ablation, intracardiac electrograms (iEGM signals), which are the result of the electrical activity of the heart muscle cells, are captured inside the heart chambers using electrodes built into catheters. The electrical activity of the heart muscle that occurs below the catheter electrodes is reflected in the iEGM signal as a short burst of high frequency activity. This is the near-field component of the iEGM signal, also known as the near-field signal (NF signal). The presence of NF signals in the iEGM signal at or inside the ablation line after RFA is an indicator of residual gaps in the ablation line, which are most often the cause of recurrent episodes of AF. In addition to NF signals, the iEGM signal can also contain far-field signals (FF signals), which are the result of electrical activity in distant parts of the heart muscle. FF signals are often similar in shape and amplitude to NF signals, and are therefore distracting components in verification of the continuity of the ablation line.
In this work, we address the problem of detection of NF signals within iEGM signals after catheter ablation for pulmonary vein isolation. The aim of the research is to develop an algorithm for detection of NF signals within iEGM signals that could assist experts in making decisions about the continuity of the ablation line. We propose an algorithm based on frequency decomposition of iEGM signals using a discrete wavelet transform. The discrete wavelet transform decomposes the iEGM input signal into several components in different frequency ranges. The iEGM signal is then reconstructed from selected individual components to isolate a specific frequency range of the iEGM input signal. Based on the amplitude characteristics of the reconstructed iEGM signal, an algorithm is used to determine whether the iEGM signal contains NF signals or not.
In this work, we first learn about the basic anatomy and physiology of the electrical conduction system in the human heart. We also present the problem of AF, the mechanisms underlying its occurrence, and catheter ablation as a method for its treatment, and describe the background of iEGM signal acquisition.
Then we present the procedure of iEGM signal acquisition procedure in 6 patients during routine catheter ablation procedures for pulmonary vein isolation in the electrophysiology laboratory of the Department of Cardiovascular Surgery, UKC Ljubljana. In addition to the signal acquisition procedure, we present the procedure for determining the gold standard for detection of NF signals within iEGM signals, which was determined based on the assessment of three independent experts on the presence of NF signals in the acquired iEGM signals. In this part of the thesis, we also describe in detail the performance of the algorithm we developed and the different variations of its performance testing. In doing so, we assess how the recognition results of our algorithm vary with the number of experts considered in the gold standard setting, and how they vary for the smaller sets of iEGM signals on which we test its performance.
The results show that our algorithm is more effective in detection of NF signals in the bipolar iEGM signals than in the unipolar iEGM signals. It achieves sensitivity of 93.6% and specificity of 90.0% in the detection of NF signals in bipolar iEGM signals. Achieved values of sensitivity and specificity of the detection of NF signals in unipolar iEGM signals were 93.8% and 70.8% respectively. The results are comparable to three other studies reported in the discussion. We also note that the recognition results with our algorithm are closer to the results of the gold standard based on consensus of at least two experts than to the gold standard represented by a single expert, but the differences are very small and probably not clinically relevant. The results of our algorithm in detection of NF signals in iEGM signals are present, which are captured in a single procedure, are comparable to the results obtained by the algorithm with the whole set of iEGM signals in 5/6 of the cases.
|
