Improved Multiscale Entropy Technique with Nearest-Neighbor Moving-Average Kernel for Nonlinear and Nonstationary Short-Time Biomedical Signal Analysis.

Analysis of biomedical signals can yield invaluable information for prognosis, diagnosis, therapy evaluation, risk assessment, and disease prevention which is often recorded as short time series data that challenges existing complexity classification algorithms such as Shannon entropy (SE) and other techniques. The purpose of this study was to improve previously developed multiscale entropy (MSE) technique by incorporating nearest-neighbor moving-average kernel, which can be used for analysis of nonlinear and non-stationary short time series physiological data. The approach was tested for robustness with respect to noise analysis using simulated sinusoidal and ECG waveforms. Feasibility of MSE to discriminate between normal sinus rhythm (NSR) and atrial fibrillation (AF) was tested on a single-lead ECG. In addition, the MSE algorithm was applied to identify pivot points of rotors that were induced in ex vivo isolated rabbit hearts. The improved MSE technique robustly estimated the complexity of the signal compared to that of SE with various noises, discriminated NSR and AF on single-lead ECG, and precisely identified the pivot points of ex vivo rotors by providing better contrast between the rotor core and the peripheral region. The improved MSE technique can provide efficient complexity analysis of variety of nonlinear and nonstationary short-time biomedical signals.

Novel approaches for quantitative electrogram analysis for intraprocedural guidance for catheter ablation: A case of a patient with persistent atrial fibrillation.

PURPOSE: Atrial fibrillation (AF) is the most common sustained cardiac arrhythmia that causes stroke affecting more than 2.3 million people in the US and is increasing in prevalence due to ageing population causing a new global epidemic. Catheter ablation with pulmonary vein isolation (PVI) to terminate AF is successful for paroxysmal AF but suffers limitations with persistent AF patients as current mapping methods cannot identify AF active substrates outside of PVI region. Recent evidences in the mechanistic understating of AF pathophysiology suggest that ectopic activity, localized re-entrant circuit with fibrillatory propagation and multiple circuit re-entries may all be involved in human AF. The authors developed novel electrogram analysis methods and validated using optical mapping data from isolated rabbit hearts to accurately identify rotor pivot points. The purpose of this study was to assess the feasibility of generating patient-specific 3D maps for intraprocedural guidance for catheter ablation using intracardiac electrograms from a persistent AF patient using novel electrogram analysis methods. METHODS: A persistent AF patient with clinical appointment for AF ablation was recruited for this study with IRB approval. 1055 electrograms throughout the left and right atrium were obtained for offline analysis with the novel approaches such as multiscale entropy, multiscale frequency, recurrence period density entropy, kurtosis and empirical mode decomposition to generate patient specific 3D maps. 3D Shannon Entropy, Renyi Entropy and Dominant frequency maps were also generated for comparison purposes along with local activation time and complex fractionated electrogram analysis maps. RESULTS: Patient specific 3D maps were obtained for each of the different approach. The 3D maps indicate potential active sites outside the PVI region. However, presence of rotors cannot be confirmed and validation of these approaches is required on a larger dataset. CONCLUSIONS: Conventional catheter mapping system can be used for generating patient specific 3D maps with short time series analysis using the novel approaches.