Non-Invasive Localization of Atrial Flutter Circuit using Recurrence Quantification Analysis and Machine Learning

Abstract : Atrial flutter presents quasi-periodic atrial activity due to circular depolarization. Given the different structure of right and left atria, spatiotemporal variability should be different. This was analyzed using recurrence quan-tification analysis. Autocorrelation signals were estimated from the unthresholded recurrence plot, calculated with a properly processed ECG to remove variability related to external sources (noise, respiratory motion, T wave overlap). Simple features were considered from the autocorre-lation that attempts to describe the atrial activity in terms of range of recurrence and periodicity. Linear classification using support vector machines and logistic regression both allowed good classification performance (max accuracy 0.8 for both). Feature selection showed that right and left AFL have significantly different cycle lengths (right vs. left: 230.63 ms vs. 206.50 ms, p < 0.01). 1. Introduction The quasi-periodic atrial activity (AA) observed on the electrocardiogram (ECG) during atrial flutter (AFL) is caused by a rotating circular depolarization of the atrium. It has been shown that beat-to-beat variability of the flutter or F waves, quantified using vectorcardiographic parameters , allowed localization of right or left atrial circuit [1]. Different variability was observed for right and left local-ization, inducing a hypothesis of varying circuit stability. With a beat-to-beat approach, instantaneous spatiotem-poral information is not preserved, which may contain information about AA. In addition, both atria are known to be remarkably different in structure. The right atrium contains many large and well-defined cardiac fibers and is relatively thin, whereas the left atrium is thick and multi-layered [2]. It is expected that spatiotemporal variability would be different. The use of recurrence quantification analysis (RQA) has been highlighted for spatiotemporal analysis and characterization of atrial fibrillation (AF) activation propagation [3, 4]. Of particular interest, atrial fibrillation recurrence behavior was characterized, and was shown to be different for recurring and non-recurring persistent AF. In this paper, RQA is employed in order to study the spatiotemporal variability related to the circular propagation of AFL activation in a non-invasive fashion. Several features are extracted from the computed recurrence signal and serves as features for classification of circuit localiza-tion. Machine learning techniques are considered in order to obtain practical classifiers as well as to understand the reason why right and left AFL are different by employing feature selection.
Complete list of metadatas

Cited literature [6 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-02285483
Contributor : Olivier Meste <>
Submitted on : Thursday, September 12, 2019 - 5:24:58 PM
Last modification on : Monday, October 14, 2019 - 6:12:49 PM

File

paperHaziq1_v1.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-02285483, version 1

Collections

Citation

Haziq Azman, Olivier Meste, Gabriel Latcu, Kushsairy Kadir. Non-Invasive Localization of Atrial Flutter Circuit using Recurrence Quantification Analysis and Machine Learning. Computing in Cardiology, Sep 2019, Singapore, Singapore. ⟨hal-02285483⟩

Share

Metrics

Record views

12

Files downloads

8