Validation of a classification system to describe fractionated electrograms in human atrial fibrillation and comparison with automated detection. (22nd September 2015)

• Record Type: Journal Article
• Authors: Hunter, RJ; Earley, MJ; Dhinoja, M; Abrams, D; Sporton, S; Schilling, RJ
• Is Part Of: Heart. Volume 95(2009)Supplement 1
• Page Start: 132
• Page End: 132
• Publication Date: 2015-09-22
• Subjects: Heart -- Diseases -- Treatment -- Periodicals; Cardiology -- Periodicals; 616.12
• Languages: English
• ISSNs: 1355-6037

Abstract

Abstract : Introduction: Fractionated electrograms (FE) are often targeted in catheter ablation of atrial fibrillation (AF) but there is no consensus as to the definition or significance of a FE. We have defined a classification of FE that can be rapidly applied by visual inspection. The aim of this study was to validate the accuracy of characterisation of FE by rapid visual inspection using this grading system and compared this against detailed manual measurement of electrogram characteristics and commercial automated FE detection. The classification system used is shown below: grade 1, near continuous fractionation (in segments lasting >1 s); grade 2, frequent but discontinuous fractionation for more than 70 ms for over 70% of recording; grade 3, as grade 2 but less than 0% of recording; grade 4, discrete complex electrograms (<70 ms with more than four direction changes); grade 5, discrete electrograms of rapid cycle length (<120 ms); grade 6, as 5 but cycle lengths greater than 120 ms. Methods: Electrograms were examined and graded by visual inspection on-line at 10 points in five patients during catheter ablation of persistent AF and the results compared with those of manual measurement determined off-line by detailed analysis blinded to the earlier grading. The temporal stability of this grading system was tested by comparing the grade determined for recordings analysed for 3–10 s. To determine consistency of a FE (and hence the time that must be analysed to assess an electrogram accurately) the percentage of the recording that was fractionated was also compared for 3–10 s. To compare our grading by visual inspection with fractionation measured by automated detection using the interval confidence level (ICL) algorithm, 2.5-s electrograms were examined at 105 points in two patients on Carto. Results: The mean grade by visual inspection was 4.6 ± 1.5. This correlated with manual measurement analysis in 86% of electrograms, with 100% correlating within 1 grade above or below. The grade determined for a 10 s sample correlated with that determined for a 3, 4, 5, 6, 7, 8, and 9 s sample in 84, 86, 90, 94, 98 and 98%, respectively. Comparing the percentage fractionation for these same time intervals to that determined for a 10 s sample showed differences of 4, 3, 3, 2, 2, 1 and 1%, respectively (none significant compared with 10 s). The ICL score by automated detection was 2.1 ± 2.1. There was minimal correlation between automated detection and visual inspection for all electrograms (r = −0.2) or comparing only grades 1–3 (r = −0.3). Conclusion: This novel classification can be quickly applied to grade electrograms by visual inspection. The grades by visual inspection correlate well with detailed manual measurement, and hence it is reasonable to grade electrograms by eye alone. The consistency of this grade over time appears relatively stable, with samples greater than 6 s correlating with 10 s in over 90%. The percentage of electrograms with continuous fractionation appears stable with little difference between 3 s and 10 s. There was no correlation between visual inspection and automated detection, suggesting that operator-selected FE are very different from those identified automatically. The clinical significance of this remains to be answered.