Detection of Abnormal Cardiac Activity Using Principal Component Analysis–-A Theoretical Study

Abstract

Electrogram-guided ablation has been recently developed for allowing better detection and localization of abnormal atrial activity that may be the source of arrhythmogeneity. Nevertheless, no clear indication for the benefit of using electrograms guided ablation over empirical ablation was established thus far, and there is a clear need of improving the localization of cardiac arrhythmogenic targets for ablation. In this paper, we propose a new approach for detection and localization of irregular cardiac activity during ablation procedures that is based on dimension reduction algorithms and principal component analysis (PCA). Using an 8×8 electrode array, our method produces manifolds that allow easy visualization and detection of possible arrhythmogenic ablation targets characterized by irregular conduction. We employ mathematical modeling and computer simulations to demonstrate the feasibility of the new approach for two well established arrhythmogenic sources for irregular conduction--spiral waves and patchy fibrosis. Our results show that the PCA method can differentiate between focal ectopic activity and spiral wave activity, as these two types of activity produce substantially different manifold shapes. Moreover, the technique allows the detection of spiral wave cores and their general meandering and drifting pattern. Fibrotic patches larger than 2 mm(2) could also be visualized using the PCA method, both for quiescent atrial tissue and for tissue exhibiting spiral wave activity. We envision that this method, contingent to further numerical and experimental validation studies in more complex, realistic geometrical configurations and with clinical data, can improve existing atrial ablation mapping capabilities, thus increasing success rates and optimizing arrhythmia management.

Cardiac spiral wave drifting due to spatial temperature gradients – A numerical study

“…Though, none of these practices were successfully implemented in clinical settings because of various limitations in their ability to accurately characterize the arrhythmogenic source zones due to noise, misleading phase and activation times that distort the reconstructed maps [11]. Moreover, novel techniques that involve more advanced signal processing methodologies to locate the pivot points of persistent rotors were proposed lately, including principal component analysis [12], and spatial Shannon entropy measurement [13]. Nevertheless, due to the yet poor understanding and the ambiguity of the correlations between the underlying arrhythmogenic activity and its electrogram manifestation, reported ablation success rates are similar for electrogram-guided as for empirical ablation procedures [4,14].…”

Doppler-Based Algorithm for Mapping Cardiac Rotors by Induced Temperature Perturbations

“…the irregularity index or CFAE) target the periphery of a source region, and may not necessarily identify the sites of the rotor tip or core [32]. Relatively new mapping methods based on signal processing (such as PCA [13], kurtosis and spatial Shannon entropy measurement [14]) are still being tested and have not been fully proven as more effective for the discovery of rotor pivot locations. Moreover, none of these mapping methods has been successfully implemented in clinical settings; hence, the need for another approach to improve mapping.…”

“…Phase analysis employing phase singularity evaluation has been used to detect rotors and their pivot points [12]. Other complex methods have been suggested to detect the origin points of rotors, including principal component analysis, multi-scale frequency (MSF), kurtosis and spatial Shannon entropy measurement [13], [14]. However, the success rate of these guided-electrogram ablation procedures has not been shown to be better than the empirical ablation procedure, which range from 60 to 90% for both [15]- [17].…”

Novel mapping techniques for rotor core detection using simulated intracardiac electrograms

“…Electrical activity in a 25 × 25 mm 2 isotropic human atrial tissue model was simulated using an extended model that incorporates both fibroblasts and myocytes, as previously published. 21 A bi-layer anatomical model was adopted, in which fibroblasts are aligned on the top and in a parallel matrix to a monolayer of myocytes. The extracellular potential, φ e , and the myocyte and fibroblast transmembrane voltages, V myo and V fib , respectively (all in [mV]), were modeled using the following bi-domain coupled equations:…”

“…The kinetic models of Courtemanche et al 25 and MacCannel et al 22 were employed for calculating I ionmyo and I ion fib , respectively, in the presence of 0.03 µM ACh as in Kneller et al 26 Rotors (minimum duration 0.4 s) were initiated via the introduction of a low concentration of ACh while preserving the con- Further, fibrotic scar tissue was introduced by increasing the concentration of fibroblasts in specific spatial sites. The patchy fibrosis was modeled based on the previous study, 21 where spatial variations in the number of fibroblasts coupled to a single myocyte were used to create the scar. The movies of a rotor in the presence of the simulated scar tissue are shown in the Supporting Information (see Video03.avi).…”