Automated algorithm-driven methods of localising drivers of persistent atrial fibrillation using atrial fibrillation cycle length and atrial fibrillation voltage

Abstract

The assessment of atrial fibrillation cycle length has played a role in the development of atrial fibrillation ablation by pulmonary vein isolation (PVI) and has also been used to assess response to ablation. Areas of rapid rotational activity in the left atrium have been implied to act as drivers of persistent atrial fibrillation and several methods have been developed to identify these potential drivers. Unprocessed atrial fibrillation electrograms show large variation in cycle length and signal amplitude. Current methods of localising driver regions rely on complex pattern recognition and subjective assessment of operators. The main hypotheses of this thesis were as follows: 1) a technique can be developed to ascertain a clinically relevant, dominant cycle length for any AF segment, 2) the automated technique, can be used to map rapid and regular activity in the left atrium, 3) a patient-tailored definition of rapid activity and low AF voltage, calculated based on patient-specific parameters is feasible; 4) paired with automated low voltage substrate analysis, dominant cycle length analysis is able to provide a framework for localising drivers of AF that is objective, transparent and requires no complex pattern recognition of subjective judgement. To test the hypotheses, a technique was developed based on manual annotation of real-world AF electrograms that was able to ascertain cycle length independent of missing segments or variable cycle length or signal amplitude. Following this, an automated algorithm was validated to determine dominant cycle length. In the following chapter, the nature of AF cycle length was investigated by investigating the patterns of rapid activity with extended AF segments and the concept of patient-tailored definitions of rapid activity was introduced. In the subsequent analysis, the effect of PVI was examined on AF voltage and the AF cycle length, focusing on rapid and regular areas and low voltage zones, and their changes. The last chapter utilised the accumulated information to test the sensitivity and specificity of a percentile-based, patient-tailored approach to low AF voltage and to present an objective, automated method of localising rapid and regular areas within low voltage zones within the left atrium. In summary, it is feasible to assess and locate rapid and regular areas, and localise low voltage zones in persistent AF with a completely automated algorithm, and patient-tailored definitions of low voltage rapid AF activity are a preferable alternative to absolute cut offs.